path: root/dynamic-layers/clang-layer/recipes-devtools/clang/files/llvm10-OpenCL-3.0-support.patch

From 31ec702cb365f4d02dd2146fb4329d642b8fc30b Mon Sep 17 00:00:00 2001
From: Naveen Saini <naveen.kumar.saini@intel.com>
Date: Wed, 7 Apr 2021 16:36:10 +0800
Subject: [PATCH 1/2] OpenCL 3.0 support

Upstream-Status: Backport [Taken from opencl-clang patches, https://github.com/intel/opencl-clang/blob/ocl-open-100/patches/clang/0001-OpenCL-3.0-support.patch]
Signed-off-by: Anton Zabaznov <anton.zabaznov@intel.com>
Signed-off-by: Naveen Saini <naveen.kumar.saini@intel.com>
---
 clang/include/clang/Basic/Builtins.def        |   65 +-
 clang/include/clang/Basic/Builtins.h          |   13 +-
 .../clang/Basic/DiagnosticParseKinds.td       |    2 +
 .../clang/Basic/DiagnosticSemaKinds.td        |    7 +
 clang/include/clang/Basic/LangOptions.def     |    2 +
 clang/include/clang/Basic/LangStandards.def   |    4 +
 .../include/clang/Basic/OpenCLExtensions.def  |   15 +
 clang/include/clang/Basic/OpenCLOptions.h     |  175 +-
 clang/include/clang/Driver/Options.td         |    2 +-
 clang/include/clang/Sema/Sema.h               |    9 +
 clang/lib/AST/ASTContext.cpp                  |    3 +-
 clang/lib/Basic/Builtins.cpp                  |   28 +-
 clang/lib/Basic/TargetInfo.cpp                |   11 +
 clang/lib/Basic/Targets.cpp                   |    1 -
 clang/lib/CodeGen/CodeGenFunction.cpp         |    6 +-
 clang/lib/Frontend/CompilerInvocation.cpp     |   22 +-
 clang/lib/Frontend/InitPreprocessor.cpp       |    6 +-
 clang/lib/Headers/opencl-c-base.h             |   75 +-
 clang/lib/Headers/opencl-c.h                  | 3228 ++++++++++++++---
 clang/lib/Parse/ParseDecl.cpp                 |   12 +-
 clang/lib/Parse/ParsePragma.cpp               |   10 +-
 clang/lib/Sema/OpenCLBuiltins.td              |   49 +-
 clang/lib/Sema/Sema.cpp                       |   47 +-
 clang/lib/Sema/SemaChecking.cpp               |   38 +-
 clang/lib/Sema/SemaDecl.cpp                   |   15 +-
 clang/lib/Sema/SemaDeclAttr.cpp               |    9 +-
 clang/lib/Sema/SemaDeclCXX.cpp                |   10 +
 clang/lib/Sema/SemaLookup.cpp                 |   19 +-
 clang/lib/Sema/SemaType.cpp                   |   14 +-
 .../CodeGenOpenCL/addr-space-struct-arg.cl    |    7 +-
 .../address-spaces-conversions.cl             |    2 +
 .../CodeGenOpenCL/address-spaces-mangling.cl  |    3 +-
 clang/test/CodeGenOpenCL/address-spaces.cl    |    4 +
 .../amdgcn-automatic-variable.cl              |    1 +
 .../CodeGenOpenCL/amdgpu-sizeof-alignof.cl    |   21 +-
 .../CodeGenOpenCL/arm-integer-dot-product.cl  |    1 +
 .../test/CodeGenOpenCL/cl-uniform-wg-size.cl  |    2 +
 clang/test/CodeGenOpenCL/fpmath.cl            |    2 +
 .../generic-address-space-feature.cl          |   28 +
 .../intel-subgroups-avc-ext-types.cl          |    1 +
 .../kernels-have-spir-cc-by-default.cl        |    3 +
 clang/test/CodeGenOpenCL/logical-ops.cl       |    1 +
 clang/test/CodeGenOpenCL/no-half.cl           |    1 +
 clang/test/CodeGenOpenCL/pipe_builtin.cl      |    3 +
 clang/test/CodeGenOpenCL/pipe_types.cl        |    1 +
 clang/test/CodeGenOpenCL/printf.cl            |    2 +
 clang/test/CodeGenOpenCL/unroll-hint.cl       |    1 +
 clang/test/Driver/autocomplete.c              |    2 +
 clang/test/Driver/opencl.cl                   |    2 +
 clang/test/Driver/unknown-std.cl              |    1 +
 clang/test/Frontend/stdlang.c                 |    1 +
 clang/test/Headers/opencl-c-header.cl         |    7 +-
 clang/test/Index/pipe-size.cl                 |    7 +
 clang/test/Preprocessor/predefined-macros.c   |   13 +
 .../Sema/feature-extensions-simult-support.cl |   75 +
 clang/test/Sema/features-ignore-pragma.cl     |   24 +
 clang/test/Sema/opencl-features-pipes.cl      |   18 +
 clang/test/Sema/opencl-features.cl            |  128 +
 clang/test/Sema/pipe_builtins_feature.cl      |   21 +
 .../address-spaces-conversions-cl2.0.cl       |    3 +
 clang/test/SemaOpenCL/address-spaces.cl       |    1 +
 .../SemaOpenCL/cl20-device-side-enqueue.cl    |   16 +-
 .../SemaOpenCL/forget-unsupported-builtins.cl |   22 +
 clang/test/SemaOpenCL/image-features.cl       |   20 +
 .../SemaOpenCL/invalid-pipe-builtin-cl2.0.cl  |    1 +
 clang/test/SemaOpenCL/storageclass-cl20.cl    |    1 +
 .../TableGen/ClangOpenCLBuiltinEmitter.cpp    |   35 +-
 67 files changed, 3656 insertions(+), 723 deletions(-)
 create mode 100644 clang/test/CodeGenOpenCL/generic-address-space-feature.cl
 create mode 100644 clang/test/Sema/feature-extensions-simult-support.cl
 create mode 100644 clang/test/Sema/features-ignore-pragma.cl
 create mode 100644 clang/test/Sema/opencl-features-pipes.cl
 create mode 100644 clang/test/Sema/opencl-features.cl
 create mode 100644 clang/test/Sema/pipe_builtins_feature.cl
 create mode 100644 clang/test/SemaOpenCL/forget-unsupported-builtins.cl
 create mode 100644 clang/test/SemaOpenCL/image-features.cl

diff --git a/clang/include/clang/Basic/Builtins.def b/clang/include/clang/Basic/Builtins.def
index 1a6c85ce2dd3..b76e277f0337 100644
--- a/clang/include/clang/Basic/Builtins.def
+++ b/clang/include/clang/Basic/Builtins.def
@@ -106,6 +106,10 @@
 #  define LANGBUILTIN(ID, TYPE, ATTRS, BUILTIN_LANG) BUILTIN(ID, TYPE, ATTRS)
 #endif
 
+#if defined(BUILTIN) && !defined(OPENCLBUILTIN)
+#  define OPENCLBUILTIN(ID, TYPE, ATTRS, BUILTIN_LANG, FEATURE) BUILTIN(ID, TYPE, ATTRS)
+#endif
+
 // Standard libc/libm functions:
 BUILTIN(__builtin_atan2 , "ddd"  , "Fne")
 BUILTIN(__builtin_atan2f, "fff"  , "Fne")
@@ -1514,50 +1518,54 @@ BUILTIN(__builtin_coro_param, "bv*v*", "n")
 
 // OpenCL v2.0 s6.13.16, s9.17.3.5 - Pipe functions.
 // We need the generic prototype, since the packet type could be anything.
-LANGBUILTIN(read_pipe, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(write_pipe, "i.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(read_pipe, "i.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
+OPENCLBUILTIN(write_pipe, "i.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
 
-LANGBUILTIN(reserve_read_pipe, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(reserve_write_pipe, "i.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(reserve_read_pipe, "i.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
+OPENCLBUILTIN(reserve_write_pipe, "i.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
 
-LANGBUILTIN(commit_write_pipe, "v.", "tn", OCLC20_LANG)
-LANGBUILTIN(commit_read_pipe, "v.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(commit_write_pipe, "v.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
+OPENCLBUILTIN(commit_read_pipe, "v.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
 
-LANGBUILTIN(sub_group_reserve_read_pipe, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(sub_group_reserve_write_pipe, "i.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(sub_group_reserve_read_pipe, "i.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
+OPENCLBUILTIN(sub_group_reserve_write_pipe, "i.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
 
-LANGBUILTIN(sub_group_commit_read_pipe, "v.", "tn", OCLC20_LANG)
-LANGBUILTIN(sub_group_commit_write_pipe, "v.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(sub_group_commit_read_pipe, "v.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
+OPENCLBUILTIN(sub_group_commit_write_pipe, "v.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
 
-LANGBUILTIN(work_group_reserve_read_pipe, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(work_group_reserve_write_pipe, "i.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(work_group_reserve_read_pipe, "i.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
+OPENCLBUILTIN(work_group_reserve_write_pipe, "i.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
 
-LANGBUILTIN(work_group_commit_read_pipe, "v.", "tn", OCLC20_LANG)
-LANGBUILTIN(work_group_commit_write_pipe, "v.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(work_group_commit_read_pipe, "v.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
+OPENCLBUILTIN(work_group_commit_write_pipe, "v.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
 
-LANGBUILTIN(get_pipe_num_packets, "Ui.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_pipe_max_packets, "Ui.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(get_pipe_num_packets, "Ui.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
+OPENCLBUILTIN(get_pipe_max_packets, "Ui.", "tn", OCLC2P_LANG, "__opencl_c_pipes")
 
 // OpenCL v2.0 s6.13.17 - Enqueue kernel functions.
 // Custom builtin check allows to perform special check of passed block arguments.
-LANGBUILTIN(enqueue_kernel, "i.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_kernel_work_group_size, "Ui.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_kernel_preferred_work_group_size_multiple, "Ui.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_kernel_max_sub_group_size_for_ndrange, "Ui.", "tn", OCLC20_LANG)
-LANGBUILTIN(get_kernel_sub_group_count_for_ndrange, "Ui.", "tn", OCLC20_LANG)
+OPENCLBUILTIN(enqueue_kernel, "i.", "tn", OCLC2P_LANG,
+              "__opencl_c_device_enqueue")
+OPENCLBUILTIN(get_kernel_work_group_size, "Ui.", "tn", OCLC2P_LANG,
+              "__opencl_c_device_enqueue")
+OPENCLBUILTIN(get_kernel_preferred_work_group_size_multiple, "Ui.", "tn",
+              OCLC2P_LANG, "__opencl_c_device_enqueue")
+OPENCLBUILTIN(get_kernel_max_sub_group_size_for_ndrange, "Ui.", "tn",
+              OCLC2P_LANG, "__opencl_c_device_enqueue")
+OPENCLBUILTIN(get_kernel_sub_group_count_for_ndrange, "Ui.", "tn", OCLC2P_LANG, "__opencl_c_device_enqueue")
 
 // OpenCL v2.0 s6.13.9 - Address space qualifier functions.
 // FIXME: Pointer parameters of OpenCL builtins should have their address space
 // requirement defined.
-LANGBUILTIN(to_global, "v*v*", "tn", OCLC20_LANG)
-LANGBUILTIN(to_local, "v*v*", "tn", OCLC20_LANG)
-LANGBUILTIN(to_private, "v*v*", "tn", OCLC20_LANG)
+OPENCLBUILTIN(to_global, "v*v*", "tn", OCLC2P_LANG, "__opencl_c_generic_address_space")
+OPENCLBUILTIN(to_local, "v*v*", "tn", OCLC2P_LANG, "__opencl_c_generic_address_space")
+OPENCLBUILTIN(to_private, "v*v*", "tn", OCLC2P_LANG, "__opencl_c_generic_address_space")
 
 // OpenCL half load/store builtin
-LANGBUILTIN(__builtin_store_half, "vdh*", "n", ALL_OCLC_LANGUAGES)
-LANGBUILTIN(__builtin_store_halff, "vfh*", "n", ALL_OCLC_LANGUAGES)
-LANGBUILTIN(__builtin_load_half, "dhC*", "nc", ALL_OCLC_LANGUAGES)
-LANGBUILTIN(__builtin_load_halff, "fhC*", "nc", ALL_OCLC_LANGUAGES)
+OPENCLBUILTIN(__builtin_store_half, "vdh*", "n", ALL_OCLC_LANGUAGES, "")
+OPENCLBUILTIN(__builtin_store_halff, "vfh*", "n", ALL_OCLC_LANGUAGES, "")
+OPENCLBUILTIN(__builtin_load_half, "dhC*", "nc", ALL_OCLC_LANGUAGES, "")
+OPENCLBUILTIN(__builtin_load_halff, "fhC*", "nc", ALL_OCLC_LANGUAGES, "")
 
 // Builtins for os_log/os_trace
 BUILTIN(__builtin_os_log_format_buffer_size, "zcC*.", "p:0:nut")
@@ -1578,3 +1586,4 @@ BUILTIN(__builtin_ms_va_copy, "vc*&c*&", "n")
 #undef BUILTIN
 #undef LIBBUILTIN
 #undef LANGBUILTIN
+#undef OPENCLBUILTIN
diff --git a/clang/include/clang/Basic/Builtins.h b/clang/include/clang/Basic/Builtins.h
index e4ed482d9068..713ea4bc267d 100644
--- a/clang/include/clang/Basic/Builtins.h
+++ b/clang/include/clang/Basic/Builtins.h
@@ -33,13 +33,13 @@ enum LanguageID {
   CXX_LANG = 0x4,     // builtin for cplusplus only.
   OBJC_LANG = 0x8,    // builtin for objective-c and objective-c++
   MS_LANG = 0x10,     // builtin requires MS mode.
-  OCLC20_LANG = 0x20, // builtin for OpenCL C 2.0 only.
+  OCLC2P_LANG = 0x20, // builtin for OpenCL C 2.0+ versions.
   OCLC1X_LANG = 0x40, // builtin for OpenCL C 1.x only.
   OMP_LANG = 0x80,    // builtin requires OpenMP.
   ALL_LANGUAGES = C_LANG | CXX_LANG | OBJC_LANG, // builtin for all languages.
   ALL_GNU_LANGUAGES = ALL_LANGUAGES | GNU_LANG,  // builtin requires GNU mode.
   ALL_MS_LANGUAGES = ALL_LANGUAGES | MS_LANG,    // builtin requires MS mode.
-  ALL_OCLC_LANGUAGES = OCLC1X_LANG | OCLC20_LANG // builtin for OCLC languages.
+  ALL_OCLC_LANGUAGES = OCLC1X_LANG | OCLC2P_LANG // builtin for OCLC languages.
 };
 
 namespace Builtin {
@@ -228,6 +228,10 @@ public:
   /// for non-builtins.
   bool canBeRedeclared(unsigned ID) const;
 
+  bool requiresFeatures(unsigned ID) const {
+    return requiresFeatures(getRecord(ID));
+  }
+
 private:
   const Info &getRecord(unsigned ID) const;
 
@@ -235,6 +239,11 @@ private:
   bool builtinIsSupported(const Builtin::Info &BuiltinInfo,
                           const LangOptions &LangOpts);
 
+  bool OclBuiltinIsSupported(const Builtin::Info &BuiltinInfo,
+                             const LangOptions &LangOpts) const;
+
+  bool requiresFeatures(const Builtin::Info &BuiltinInfo) const;
+
   /// Helper function for isPrintfLike and isScanfLike.
   bool isLike(unsigned ID, unsigned &FormatIdx, bool &HasVAListArg,
               const char *Fmt) const;
diff --git a/clang/include/clang/Basic/DiagnosticParseKinds.td b/clang/include/clang/Basic/DiagnosticParseKinds.td
index 33adf093693f..43ea63586f91 100644
--- a/clang/include/clang/Basic/DiagnosticParseKinds.td
+++ b/clang/include/clang/Basic/DiagnosticParseKinds.td
@@ -1199,6 +1199,8 @@ def warn_pragma_unsupported_extension : Warning<
   "unsupported OpenCL extension %0 - ignoring">, InGroup<IgnoredPragmas>;
 def warn_pragma_extension_is_core : Warning<
   "OpenCL extension %0 is core feature or supported optional core feature - ignoring">, InGroup<DiagGroup<"pedantic-core-features">>, DefaultIgnore;
+def warn_opencl_pragma_feature_ignore : Warning<
+  "OpenCL feature support can't be controlled via pragma, ignoring">, InGroup<IgnoredPragmas>;
 
 // OpenCL errors.
 def err_opencl_taking_function_address_parser : Error<
diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index 917377420505..91648aa27820 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -9378,6 +9378,13 @@ def ext_opencl_ext_vector_type_rgba_selector: ExtWarn<
 def err_openclcxx_placement_new : Error<
   "use of placement new requires explicit declaration">;
 
+def err_opencl_pipes_require_feat : Error<
+  "usage of OpenCL pipes requires feature support">;
+def err_opencl_memory_scope_require_feat : Error<
+  "usage of memory scope requires feature support">;
+def err_opencl_memory_ordering_require_feat : Error<
+  "usage of memory ordering requires feature support">;
+
 // MIG routine annotations.
 def warn_mig_server_routine_does_not_return_kern_return_t : Warning<
   "'mig_server_routine' attribute only applies to routines that return a kern_return_t">,
diff --git a/clang/include/clang/Basic/LangOptions.def b/clang/include/clang/Basic/LangOptions.def
index 3319a3123976..406f45c0be69 100644
--- a/clang/include/clang/Basic/LangOptions.def
+++ b/clang/include/clang/Basic/LangOptions.def
@@ -203,6 +203,8 @@ LANGOPT(OpenCL            , 1, 0, "OpenCL")
 LANGOPT(OpenCLVersion     , 32, 0, "OpenCL C version")
 LANGOPT(OpenCLCPlusPlus   , 1, 0, "C++ for OpenCL")
 LANGOPT(OpenCLCPlusPlusVersion     , 32, 0, "C++ for OpenCL version")
+LANGOPT(OpenCLGenericKeyword       , 1,  0, "OpenCL generic keyword")
+LANGOPT(OpenCLPipeKeyword       , 1,  0, "OpenCL pipe keyword")
 LANGOPT(NativeHalfType    , 1, 0, "Native half type support")
 LANGOPT(NativeHalfArgsAndReturns, 1, 0, "Native half args and returns")
 LANGOPT(HalfArgsAndReturns, 1, 0, "half args and returns")
diff --git a/clang/include/clang/Basic/LangStandards.def b/clang/include/clang/Basic/LangStandards.def
index 7f1a24db7e9b..69aaba3ff5a2 100644
--- a/clang/include/clang/Basic/LangStandards.def
+++ b/clang/include/clang/Basic/LangStandards.def
@@ -167,6 +167,9 @@ LANGSTANDARD(opencl12, "cl1.2",
 LANGSTANDARD(opencl20, "cl2.0",
              OpenCL, "OpenCL 2.0",
              LineComment | C99 | Digraphs | HexFloat | OpenCL)
+LANGSTANDARD(opencl30, "cl3.0",
+             OpenCL, "OpenCL 3.0",
+             LineComment | C99 | Digraphs | HexFloat | OpenCL)
 LANGSTANDARD(openclcpp, "clc++",
              OpenCL, "C++ for OpenCL",
              LineComment | CPlusPlus | CPlusPlus11 | CPlusPlus14 | CPlusPlus17 |
@@ -176,6 +179,7 @@ LANGSTANDARD_ALIAS_DEPR(opencl10, "CL")
 LANGSTANDARD_ALIAS_DEPR(opencl11, "CL1.1")
 LANGSTANDARD_ALIAS_DEPR(opencl12, "CL1.2")
 LANGSTANDARD_ALIAS_DEPR(opencl20, "CL2.0")
+LANGSTANDARD_ALIAS_DEPR(opencl30, "CL3.0")
 LANGSTANDARD_ALIAS_DEPR(openclcpp, "CLC++")
 
 // CUDA
diff --git a/clang/include/clang/Basic/OpenCLExtensions.def b/clang/include/clang/Basic/OpenCLExtensions.def
index 517481584313..608f78a13eef 100644
--- a/clang/include/clang/Basic/OpenCLExtensions.def
+++ b/clang/include/clang/Basic/OpenCLExtensions.def
@@ -93,6 +93,21 @@ OPENCLEXT_INTERNAL(cl_intel_subgroups, 120, ~0U)
 OPENCLEXT_INTERNAL(cl_intel_subgroups_short, 120, ~0U)
 OPENCLEXT_INTERNAL(cl_intel_device_side_avc_motion_estimation, 120, ~0U)
 
+OPENCLEXT_INTERNAL(__opencl_c_pipes, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_generic_address_space, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_work_group_collective_functions, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_atomic_order_acq_rel, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_atomic_order_seq_cst, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_atomic_scope_device, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_atomic_scope_all_devices, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_subgroups, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_3d_image_writes, 100, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_device_enqueue, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_read_write_images, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_program_scope_global_variables, 200, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_fp64, 120, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_int64, 100, ~0U)
+OPENCLEXT_INTERNAL(__opencl_c_images, 100, ~0U)
 #undef OPENCLEXT_INTERNAL
 
 #ifdef OPENCLEXT
diff --git a/clang/include/clang/Basic/OpenCLOptions.h b/clang/include/clang/Basic/OpenCLOptions.h
index 15661154eab5..9a3a8e33cadd 100644
--- a/clang/include/clang/Basic/OpenCLOptions.h
+++ b/clang/include/clang/Basic/OpenCLOptions.h
@@ -16,11 +16,16 @@
 
 #include "clang/Basic/LangOptions.h"
 #include "llvm/ADT/StringMap.h"
+#include "llvm/ADT/StringSwitch.h"
 
 namespace clang {
 
 /// OpenCL supported extensions and optional core features
 class OpenCLOptions {
+  // OpenCL Version
+  unsigned CLVer = 120;
+  bool IsOpenCLCPlusPlus = false;
+
   struct Info {
     bool Supported; // Is this option supported
     bool Enabled;   // Is this option enabled
@@ -31,7 +36,38 @@ class OpenCLOptions {
       :Supported(S), Enabled(E), Avail(A), Core(C){}
   };
   llvm::StringMap<Info> OptMap;
+
 public:
+  void setOpenCLVersion(const LangOptions &LO) {
+    IsOpenCLCPlusPlus = LO.OpenCLCPlusPlus;
+    CLVer = IsOpenCLCPlusPlus ? 200 : LO.OpenCLVersion;
+  }
+
+  // Get extension which is semantically equivalent to a given feature
+  // if exists (e.g. __opencl_c_subgroups -> cl_khr_subgroups)
+  llvm::Optional<StringRef> getEquivalentExtension(StringRef Feature) const {
+    return llvm::StringSwitch<llvm::Optional<StringRef>>(Feature)
+        .Case("__opencl_c_3d_image_writes",
+              Optional<StringRef>("cl_khr_3d_image_writes"))
+        .Case("__opencl_c_subgroups", Optional<StringRef>("cl_khr_subgroups"))
+        .Case("__opencl_c_fp64", Optional<StringRef>("cl_khr_fp64"))
+        .Default(Optional<StringRef>());
+  }
+
+  // Same as above but for extensions
+  llvm::Optional<StringRef> getEquivalentFeature(StringRef Extension) const {
+    return llvm::StringSwitch<llvm::Optional<StringRef>>(Extension)
+        .Case("cl_khr_3d_image_writes",
+              Optional<StringRef>("__opencl_c_3d_image_writes"))
+        .Case("cl_khr_subgroups", Optional<StringRef>("__opencl_c_subgroups"))
+        .Case("cl_khr_fp64", Optional<StringRef>("__opencl_c_fp64"))
+        .Default(Optional<StringRef>());
+  }
+
+  bool isFeature(llvm::StringRef Ext) const {
+    return Ext.startswith("__opencl_c");
+  }
+
   bool isKnown(llvm::StringRef Ext) const {
     return OptMap.find(Ext) != OptMap.end();
   }
@@ -42,32 +78,88 @@ public:
 
   // Is supported as either an extension or an (optional) core feature for
   // OpenCL version \p CLVer.
-  bool isSupported(llvm::StringRef Ext, const LangOptions &LO) const {
+  bool isSupported(llvm::StringRef Ext) const {
     // In C++ mode all extensions should work at least as in v2.0.
-    auto CLVer = LO.OpenCLCPlusPlus ? 200 : LO.OpenCLVersion;
     auto I = OptMap.find(Ext)->getValue();
     return I.Supported && I.Avail <= CLVer;
   }
 
   // Is supported (optional) OpenCL core features for OpenCL version \p CLVer.
   // For supported extension, return false.
-  bool isSupportedCore(llvm::StringRef Ext, const LangOptions &LO) const {
+  bool isSupportedCore(llvm::StringRef Ext) const {
     // In C++ mode all extensions should work at least as in v2.0.
-    auto CLVer = LO.OpenCLCPlusPlus ? 200 : LO.OpenCLVersion;
     auto I = OptMap.find(Ext)->getValue();
     return I.Supported && I.Avail <= CLVer && I.Core != ~0U && CLVer >= I.Core;
   }
 
   // Is supported OpenCL extension for OpenCL version \p CLVer.
   // For supported (optional) core feature, return false.
-  bool isSupportedExtension(llvm::StringRef Ext, const LangOptions &LO) const {
+  bool isSupportedExtension(llvm::StringRef Ext) const {
     // In C++ mode all extensions should work at least as in v2.0.
-    auto CLVer = LO.OpenCLCPlusPlus ? 200 : LO.OpenCLVersion;
     auto I = OptMap.find(Ext)->getValue();
     return I.Supported && I.Avail <= CLVer && (I.Core == ~0U || CLVer < I.Core);
   }
 
+  // Support features whose support is directly related to the
+  // specific OpenCL version. For example, OpenCL 2.0 supports
+  // all features that are optional in 3.0
+  void adjustFeatures() {
+    // Support int64 by default (assume compiling for FULL profile)
+    OptMap["__opencl_c_int64"].Supported = true;
+
+    if (CLVer >= 300) {
+      // Simultaneously support extension and corresponding feature
+      for (llvm::StringRef F :
+           {"__opencl_c_subgroups", "__opencl_c_3d_image_writes",
+            "__opencl_c_fp64"}) {
+        auto Ext = getEquivalentExtension(F);
+        OptMap[*Ext].Supported = OptMap[F].Supported;
+      }
+
+      // OpenCL C compilers that define the feature macro __opencl_c_pipes or
+      // or __opencl_c_device_enqueue must also define the
+      // feature macro __opencl_c_generic_address_space.
+      if (OptMap["__opencl_c_pipes"].Supported ||
+          OptMap["__opencl_c_device_enqueue"].Supported)
+        OptMap["__opencl_c_generic_address_space"].Supported = true;
+
+      // OpenCL C compilers that define the feature macro
+      // __opencl_c_3d_image_writes or __opencl_c_read_write_images must also
+      // define the feature macro __opencl_c_images
+      if (OptMap["__opencl_c_3d_image_writes"].Supported ||
+          OptMap["__opencl_c_read_write_images"].Supported)
+        OptMap["__opencl_c_images"].Supported = true;
+
+      // All other features are already supported with options
+      // or in target settings
+      return;
+    }
+
+    auto FeaturesRange = llvm::make_filter_range(
+        OptMap, [&](llvm::StringMapEntry<Info> &OptVal) {
+          auto Opt = OptVal.getKey();
+          return isFeature(Opt);
+        });
+
+    for (auto &It : FeaturesRange) {
+      auto &Info = It.getValue();
+      // For OpenCL version less then 3.0 some
+      // features should be supported simulateneously
+      // with specific extension
+      if (Optional<llvm::StringRef> Ext = getEquivalentExtension(It.getKey()))
+        Info.Supported = Info.Enabled = OptMap[*Ext].Supported;
+      else if (Info.Avail <= CLVer)
+        Info.Supported = Info.Enabled = true;
+    }
+  }
+
   void enable(llvm::StringRef Ext, bool V = true) {
+    // Ignore disabling extensions if corresponding features
+    // already supported for OpenCL version higher then 3.0
+    if (CLVer >= 300)
+      if (Optional<llvm::StringRef> F = getEquivalentFeature(Ext))
+        if (V != OptMap[*F].Enabled)
+          return;
     OptMap[Ext].Enabled = V;
   }
 
@@ -96,7 +188,7 @@ public:
     OptMap[Ext].Supported = V;
   }
 
-  OpenCLOptions(){
+  OpenCLOptions() {
 #define OPENCLEXT_INTERNAL(Ext, AvailVer, CoreVer) \
     OptMap[#Ext].Avail = AvailVer; \
     OptMap[#Ext].Core = CoreVer;
@@ -104,35 +196,86 @@ public:
   }
 
   void addSupport(const OpenCLOptions &Opts) {
+    assert(IsOpenCLCPlusPlus == Opts.IsOpenCLCPlusPlus && CLVer == Opts.CLVer);
     for (auto &I:Opts.OptMap)
-      if (I.second.Supported)
+      if (I.second.Supported) {
         OptMap[I.getKey()].Supported = true;
+        // All features are enabled as they are supported
+        if (isFeature(I.getKey()))
+          OptMap[I.getKey()].Enabled = true;
+      }
+    if (CLVer >= 300) {
+      // Enabling extensions with respect to features
+      for (llvm::StringRef Ext :
+           {"cl_khr_3d_image_writes", "cl_khr_subgroups", "cl_khr_fp64"}) {
+        auto Feature = getEquivalentFeature(Ext);
+        enable(Ext, OptMap[*Feature].Enabled);
+      }
+    }
   }
 
   void copy(const OpenCLOptions &Opts) {
+    CLVer = Opts.CLVer;
+    IsOpenCLCPlusPlus = Opts.IsOpenCLCPlusPlus;
     OptMap = Opts.OptMap;
   }
 
   // Turn on or off support of all options.
   void supportAll(bool On = true) {
-    for (llvm::StringMap<Info>::iterator I = OptMap.begin(),
-         E = OptMap.end(); I != E; ++I)
-      I->second.Supported = On;
+    for (llvm::StringMap<Info>::iterator I = OptMap.begin(), E = OptMap.end();
+         I != E; ++I)
+      if (!isFeature(I->getKey()))
+        I->second.Supported = On;
   }
 
   void disableAll() {
-    for (llvm::StringMap<Info>::iterator I = OptMap.begin(),
-         E = OptMap.end(); I != E; ++I)
-      I->second.Enabled = false;
+    for (llvm::StringMap<Info>::iterator I = OptMap.begin(), E = OptMap.end();
+         I != E; ++I) {
+      auto Ext = I->getKey();
+      if (!isFeature(Ext))
+        enable(Ext, false);
+    }
   }
 
-  void enableSupportedCore(LangOptions LO) {
+  void enableSupportedCore() {
     for (llvm::StringMap<Info>::iterator I = OptMap.begin(), E = OptMap.end();
          I != E; ++I)
-      if (isSupportedCore(I->getKey(), LO))
+      if (isSupportedCore(I->getKey()))
         I->second.Enabled = true;
   }
 
+  // This enum specifies how OpenCL versions map into values
+  // for encoding. This is used when generating built-ins
+  // from tablegen
+  enum OpenCLVersionsEncodings : unsigned short {
+    OPENCL_C_100_CODE = 0x1,
+    OPENCL_C_110_CODE = 0x2,
+    OPENCL_C_120_CODE = 0x4,
+    OPENCL_C_200_CODE = 0x8,
+    OPENCL_C_300_CODE = 0x10,
+    OPENCL_C_ALL_CODE = 0x1f
+  };
+
+  // Encode version into single integer
+  static unsigned short EncodeVersion(unsigned OpenCLVersion) {
+    switch (OpenCLVersion) {
+    default:
+      llvm_unreachable("Unknown OpenCL version");
+    case 0:
+      return OpenCLVersionsEncodings::OPENCL_C_ALL_CODE;
+    case 100:
+      return OpenCLVersionsEncodings::OPENCL_C_100_CODE;
+    case 110:
+      return OpenCLVersionsEncodings::OPENCL_C_110_CODE;
+    case 120:
+      return OpenCLVersionsEncodings::OPENCL_C_120_CODE;
+    case 200:
+      return OpenCLVersionsEncodings::OPENCL_C_200_CODE;
+    case 300:
+      return OpenCLVersionsEncodings::OPENCL_C_300_CODE;
+    }
+  }
+
   friend class ASTWriter;
   friend class ASTReader;
 };
diff --git a/clang/include/clang/Driver/Options.td b/clang/include/clang/Driver/Options.td
index 391c895a453b..e03c22c749ad 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -522,7 +522,7 @@ def cl_mad_enable : Flag<["-"], "cl-mad-enable">, Group<opencl_Group>, Flags<[CC
 def cl_no_signed_zeros : Flag<["-"], "cl-no-signed-zeros">, Group<opencl_Group>, Flags<[CC1Option]>,
   HelpText<"OpenCL only. Allow use of less precise no signed zeros computations in the generated binary.">;
 def cl_std_EQ : Joined<["-"], "cl-std=">, Group<opencl_Group>, Flags<[CC1Option]>,
-  HelpText<"OpenCL language standard to compile for.">, Values<"cl,CL,cl1.1,CL1.1,cl1.2,CL1.2,cl2.0,CL2.0,clc++,CLC++">;
+  HelpText<"OpenCL language standard to compile for.">, Values<"cl,CL,cl1.1,CL1.1,cl1.2,CL1.2,cl2.0,CL2.0,cl3.0,CL3.0,clc++,CLC++">;
 def cl_denorms_are_zero : Flag<["-"], "cl-denorms-are-zero">, Group<opencl_Group>, Flags<[CC1Option]>,
   HelpText<"OpenCL only. Allow denormals to be flushed to zero.">;
 def cl_fp32_correctly_rounded_divide_sqrt : Flag<["-"], "cl-fp32-correctly-rounded-divide-sqrt">, Group<opencl_Group>, Flags<[CC1Option]>,
diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index 842e49602274..d8ac91bc8a31 100644
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -9632,6 +9632,10 @@ public:
   /// \return true if type is disabled.
   bool checkOpenCLDisabledDecl(const NamedDecl &D, const Expr &E);
 
+  bool checkOpenCLSubgroupExtForCallExpr(CallExpr *Call);
+
+  bool isSupportedOpenCLOMemoryOrdering(int64_t Ordering) const;
+
   //===--------------------------------------------------------------------===//
   // OpenMP directives and clauses.
   //
@@ -11102,6 +11106,11 @@ public:
   /// that the user intended an assignment used as condition.
   void DiagnoseEqualityWithExtraParens(ParenExpr *ParenE);
 
+  template <typename DiagLocT, typename DiagInfoT>
+  void DiagnoseOpenCLRequiresOption(llvm::StringRef OpenCLOptName,
+                                    DiagLocT DiagLoc, DiagInfoT DiagInfo,
+                                    unsigned Selector, SourceRange SrcRange);
+
   /// CheckCXXBooleanCondition - Returns true if conversion to bool is invalid.
   ExprResult CheckCXXBooleanCondition(Expr *CondExpr, bool IsConstexpr = false);
 
diff --git a/clang/lib/AST/ASTContext.cpp b/clang/lib/AST/ASTContext.cpp
index 1be72efe4de8..2a81fbcde79d 100644
--- a/clang/lib/AST/ASTContext.cpp
+++ b/clang/lib/AST/ASTContext.cpp
@@ -1490,7 +1490,8 @@ void ASTContext::InitBuiltinTypes(const TargetInfo &Target,
   ObjCSuperType = QualType();
 
   // void * type
-  if (LangOpts.OpenCLVersion >= 200) {
+  if (Target.getSupportedOpenCLOpts().isEnabled(
+          "__opencl_c_generic_address_space")) {
     auto Q = VoidTy.getQualifiers();
     Q.setAddressSpace(LangAS::opencl_generic);
     VoidPtrTy = getPointerType(getCanonicalType(
diff --git a/clang/lib/Basic/Builtins.cpp b/clang/lib/Basic/Builtins.cpp
index 0cd89df41b67..7a3067345098 100644
--- a/clang/lib/Basic/Builtins.cpp
+++ b/clang/lib/Basic/Builtins.cpp
@@ -23,6 +23,8 @@ static const Builtin::Info BuiltinInfo[] = {
   { #ID, TYPE, ATTRS, nullptr, ALL_LANGUAGES, nullptr },
 #define LANGBUILTIN(ID, TYPE, ATTRS, LANGS)                                    \
   { #ID, TYPE, ATTRS, nullptr, LANGS, nullptr },
+#define OPENCLBUILTIN(ID, TYPE, ATTRS, LANGS, FEATURE)                         \
+  {#ID, TYPE, ATTRS, nullptr, LANGS, FEATURE},
 #define LIBBUILTIN(ID, TYPE, ATTRS, HEADER, LANGS)                             \
   { #ID, TYPE, ATTRS, HEADER, LANGS, nullptr },
 #include "clang/Basic/Builtins.def"
@@ -69,16 +71,17 @@ bool Builtin::Context::builtinIsSupported(const Builtin::Info &BuiltinInfo,
   bool ObjCUnsupported = !LangOpts.ObjC && BuiltinInfo.Langs == OBJC_LANG;
   bool OclC1Unsupported = (LangOpts.OpenCLVersion / 100) != 1 &&
                           (BuiltinInfo.Langs & ALL_OCLC_LANGUAGES ) ==  OCLC1X_LANG;
-  bool OclC2Unsupported =
-      (LangOpts.OpenCLVersion != 200 && !LangOpts.OpenCLCPlusPlus) &&
-      (BuiltinInfo.Langs & ALL_OCLC_LANGUAGES) == OCLC20_LANG;
+  bool OclC2PUnsupported =
+      (BuiltinInfo.Langs & ALL_OCLC_LANGUAGES) == OCLC2P_LANG &&
+      ((LangOpts.OpenCLVersion < 200 && !LangOpts.OpenCLCPlusPlus) ||
+       !OclBuiltinIsSupported(BuiltinInfo, LangOpts));
   bool OclCUnsupported = !LangOpts.OpenCL &&
                          (BuiltinInfo.Langs & ALL_OCLC_LANGUAGES);
   bool OpenMPUnsupported = !LangOpts.OpenMP && BuiltinInfo.Langs == OMP_LANG;
   bool CPlusPlusUnsupported =
       !LangOpts.CPlusPlus && BuiltinInfo.Langs == CXX_LANG;
   return !BuiltinsUnsupported && !MathBuiltinsUnsupported && !OclCUnsupported &&
-         !OclC1Unsupported && !OclC2Unsupported && !OpenMPUnsupported &&
+         !OclC1Unsupported && !OclC2PUnsupported && !OpenMPUnsupported &&
          !GnuModeUnsupported && !MSModeUnsupported && !ObjCUnsupported &&
          !CPlusPlusUnsupported;
 }
@@ -191,3 +194,20 @@ bool Builtin::Context::canBeRedeclared(unsigned ID) const {
          (!hasReferenceArgsOrResult(ID) &&
           !hasCustomTypechecking(ID));
 }
+
+bool Builtin::Context::OclBuiltinIsSupported(
+    const Builtin::Info &BuiltinInfo, const LangOptions &LangOpts) const {
+  if (!requiresFeatures(BuiltinInfo))
+    return true;
+
+  return llvm::StringSwitch<bool>(BuiltinInfo.Features)
+      .Case("__opencl_c_device_enqueue", LangOpts.Blocks)
+      .Case("__opencl_c_generic_address_space", LangOpts.OpenCLGenericKeyword)
+      .Case("__opencl_c_pipes", LangOpts.OpenCLPipeKeyword)
+      .Default(false);
+}
+
+bool Builtin::Context::requiresFeatures(
+    const Builtin::Info &BuiltinInfo) const {
+  return BuiltinInfo.Features && llvm::StringRef(BuiltinInfo.Features) != "";
+}
diff --git a/clang/lib/Basic/TargetInfo.cpp b/clang/lib/Basic/TargetInfo.cpp
index 3a21a19e1f19..795311d5934e 100644
--- a/clang/lib/Basic/TargetInfo.cpp
+++ b/clang/lib/Basic/TargetInfo.cpp
@@ -377,6 +377,17 @@ void TargetInfo::adjust(LangOptions &Opts) {
     HalfFormat = &llvm::APFloat::IEEEhalf();
     FloatFormat = &llvm::APFloat::IEEEsingle();
     LongDoubleFormat = &llvm::APFloat::IEEEquad();
+
+    auto &SupportedOCLOpts = getTargetOpts().SupportedOpenCLOptions;
+
+    SupportedOCLOpts.setOpenCLVersion(Opts);
+    SupportedOCLOpts.adjustFeatures();
+
+    if (!Opts.OpenCLCPlusPlus && Opts.OpenCLVersion >= 200)
+      Opts.Blocks = SupportedOCLOpts.isSupported("__opencl_c_device_enqueue");
+    Opts.OpenCLGenericKeyword =
+        SupportedOCLOpts.isSupported("__opencl_c_generic_address_space");
+    Opts.OpenCLPipeKeyword = SupportedOCLOpts.isSupported("__opencl_c_pipes");
   }
 
   if (Opts.LongDoubleSize) {
diff --git a/clang/lib/Basic/Targets.cpp b/clang/lib/Basic/Targets.cpp
index c063f8ca4472..b394be18a234 100644
--- a/clang/lib/Basic/Targets.cpp
+++ b/clang/lib/Basic/Targets.cpp
@@ -39,7 +39,6 @@
 #include "clang/Basic/Diagnostic.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/Triple.h"
-
 using namespace clang;
 
 namespace clang {
diff --git a/clang/lib/CodeGen/CodeGenFunction.cpp b/clang/lib/CodeGen/CodeGenFunction.cpp
index 648e6d9c214a..6d839fa61abc 100644
--- a/clang/lib/CodeGen/CodeGenFunction.cpp
+++ b/clang/lib/CodeGen/CodeGenFunction.cpp
@@ -2303,11 +2303,11 @@ void CodeGenFunction::checkTargetFeatures(SourceLocation Loc,
   std::string MissingFeature;
   if (BuiltinID) {
     SmallVector<StringRef, 1> ReqFeatures;
-    const char *FeatureList =
-        CGM.getContext().BuiltinInfo.getRequiredFeatures(BuiltinID);
     // Return if the builtin doesn't have any required features.
-    if (!FeatureList || StringRef(FeatureList) == "")
+    if (!CGM.getContext().BuiltinInfo.requiresFeatures(BuiltinID))
       return;
+    const char *FeatureList =
+        CGM.getContext().BuiltinInfo.getRequiredFeatures(BuiltinID);
     StringRef(FeatureList).split(ReqFeatures, ',');
     if (!hasRequiredFeatures(ReqFeatures, CGM, FD, MissingFeature))
       CGM.getDiags().Report(Loc, diag::err_builtin_needs_feature)
diff --git a/clang/lib/Frontend/CompilerInvocation.cpp b/clang/lib/Frontend/CompilerInvocation.cpp
index e98a407ac42f..18fa06bf3c6d 100644
--- a/clang/lib/Frontend/CompilerInvocation.cpp
+++ b/clang/lib/Frontend/CompilerInvocation.cpp
@@ -2298,6 +2298,8 @@ void CompilerInvocation::setLangDefaults(LangOptions &Opts, InputKind IK,
     Opts.OpenCLVersion = 120;
   else if (LangStd == LangStandard::lang_opencl20)
     Opts.OpenCLVersion = 200;
+  else if (LangStd == LangStandard::lang_opencl30)
+    Opts.OpenCLVersion = 300;
   else if (LangStd == LangStandard::lang_openclcpp)
     Opts.OpenCLCPlusPlusVersion = 100;
 
@@ -2498,14 +2500,15 @@ static void ParseLangArgs(LangOptions &Opts, ArgList &Args, InputKind IK,
   // -cl-std only applies for OpenCL language standards.
   // Override the -std option in this case.
   if (const Arg *A = Args.getLastArg(OPT_cl_std_EQ)) {
-    LangStandard::Kind OpenCLLangStd
-      = llvm::StringSwitch<LangStandard::Kind>(A->getValue())
-        .Cases("cl", "CL", LangStandard::lang_opencl10)
-        .Cases("cl1.1", "CL1.1", LangStandard::lang_opencl11)
-        .Cases("cl1.2", "CL1.2", LangStandard::lang_opencl12)
-        .Cases("cl2.0", "CL2.0", LangStandard::lang_opencl20)
-        .Cases("clc++", "CLC++", LangStandard::lang_openclcpp)
-        .Default(LangStandard::lang_unspecified);
+    LangStandard::Kind OpenCLLangStd =
+        llvm::StringSwitch<LangStandard::Kind>(A->getValue())
+            .Cases("cl", "CL", LangStandard::lang_opencl10)
+            .Cases("cl1.1", "CL1.1", LangStandard::lang_opencl11)
+            .Cases("cl1.2", "CL1.2", LangStandard::lang_opencl12)
+            .Cases("cl2.0", "CL2.0", LangStandard::lang_opencl20)
+            .Cases("cl3.0", "CL3.0", LangStandard::lang_opencl30)
+            .Cases("clc++", "CLC++", LangStandard::lang_openclcpp)
+            .Default(LangStandard::lang_unspecified);
 
     if (OpenCLLangStd == LangStandard::lang_unspecified) {
       Diags.Report(diag::err_drv_invalid_value)
@@ -2787,8 +2790,7 @@ static void ParseLangArgs(LangOptions &Opts, ArgList &Args, InputKind IK,
 
   Opts.RTTI = Opts.CPlusPlus && !Args.hasArg(OPT_fno_rtti);
   Opts.RTTIData = Opts.RTTI && !Args.hasArg(OPT_fno_rtti_data);
-  Opts.Blocks = Args.hasArg(OPT_fblocks) || (Opts.OpenCL
-    && Opts.OpenCLVersion == 200);
+  Opts.Blocks = Args.hasArg(OPT_fblocks);
   Opts.BlocksRuntimeOptional = Args.hasArg(OPT_fblocks_runtime_optional);
   Opts.Coroutines = Opts.CPlusPlus2a || Args.hasArg(OPT_fcoroutines_ts);
 
diff --git a/clang/lib/Frontend/InitPreprocessor.cpp b/clang/lib/Frontend/InitPreprocessor.cpp
index c273cb96d9b9..aefd208e6cd3 100644
--- a/clang/lib/Frontend/InitPreprocessor.cpp
+++ b/clang/lib/Frontend/InitPreprocessor.cpp
@@ -445,6 +445,9 @@ static void InitializeStandardPredefinedMacros(const TargetInfo &TI,
       case 200:
         Builder.defineMacro("__OPENCL_C_VERSION__", "200");
         break;
+      case 300:
+        Builder.defineMacro("__OPENCL_C_VERSION__", "300");
+        break;
       default:
         llvm_unreachable("Unsupported OpenCL version");
       }
@@ -453,6 +456,7 @@ static void InitializeStandardPredefinedMacros(const TargetInfo &TI,
     Builder.defineMacro("CL_VERSION_1_1", "110");
     Builder.defineMacro("CL_VERSION_1_2", "120");
     Builder.defineMacro("CL_VERSION_2_0", "200");
+    Builder.defineMacro("CL_VERSION_3_0", "300");
 
     if (TI.isLittleEndian())
       Builder.defineMacro("__ENDIAN_LITTLE__");
@@ -1101,7 +1105,7 @@ static void InitializePredefinedMacros(const TargetInfo &TI,
   // OpenCL definitions.
   if (LangOpts.OpenCL) {
 #define OPENCLEXT(Ext)                                                         \
-  if (TI.getSupportedOpenCLOpts().isSupported(#Ext, LangOpts))                 \
+  if (TI.getSupportedOpenCLOpts().isSupported(#Ext))                           \
     Builder.defineMacro(#Ext);
 #include "clang/Basic/OpenCLExtensions.def"
 
diff --git a/clang/lib/Headers/opencl-c-base.h b/clang/lib/Headers/opencl-c-base.h
index 430e07d36f62..2cc688ccc3da 100644
--- a/clang/lib/Headers/opencl-c-base.h
+++ b/clang/lib/Headers/opencl-c-base.h
@@ -9,6 +9,59 @@
 #ifndef _OPENCL_BASE_H_
 #define _OPENCL_BASE_H_
 
+// Add predefined macros to build headers with standalone executable
+#ifndef CL_VERSION_3_0
+  #define CL_VERSION_3_0 300
+#endif
+
+// Define features for 2.0 for header backward compatibility
+#ifndef __opencl_c_int64
+  #define __opencl_c_int64 1
+#endif
+#if __OPENCL_C_VERSION__ != CL_VERSION_3_0
+  #ifndef __opencl_c_images
+    #define __opencl_c_images 1
+  #endif
+#endif
+#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ == CL_VERSION_2_0)
+#ifndef __opencl_c_pipes
+  #define __opencl_c_pipes 1
+#endif
+#ifndef __opencl_c_generic_address_space
+  #define __opencl_c_generic_address_space 1
+#endif
+#ifndef __opencl_c_work_group_collective_functions
+  #define __opencl_c_work_group_collective_functions 1
+#endif
+#ifndef __opencl_c_atomic_order_acq_rel
+  #define __opencl_c_atomic_order_acq_rel 1
+#endif
+#ifndef __opencl_c_atomic_order_seq_cst
+  #define __opencl_c_atomic_order_seq_cst 1
+#endif
+#ifndef __opencl_c_atomic_scope_device
+ #define __opencl_c_atomic_scope_device 1
+#endif
+#ifndef __opencl_c_atomic_scope_all_devices
+  #define __opencl_c_atomic_scope_all_devices 1
+#endif
+#ifndef __opencl_c_subgroups
+  #define __opencl_c_subgroups 1
+#endif
+#ifndef __opencl_c_3d_image_writes
+  #define __opencl_c_3d_image_writes 1
+#endif
+#ifndef __opencl_c_device_enqueue
+  #define __opencl_c_device_enqueue 1
+#endif
+#ifndef __opencl_c_read_write_images
+  #define __opencl_c_read_write_images 1
+#endif
+#ifndef __opencl_c_program_scope_global_variables
+  #define __opencl_c_program_scope_global_variables 1
+#endif
+#endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ == CL_VERSION_2_0)
+
 // built-in scalar data types:
 
 /**
@@ -115,7 +168,12 @@ typedef half half4 __attribute__((ext_vector_type(4)));
 typedef half half8 __attribute__((ext_vector_type(8)));
 typedef half half16 __attribute__((ext_vector_type(16)));
 #endif
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+
+#ifndef __opencl_c_fp64
+  #define __opencl_c_fp64 1
+#endif
+
 #if __OPENCL_C_VERSION__ < CL_VERSION_1_2
 #pragma OPENCL EXTENSION cl_khr_fp64 : enable
 #endif
@@ -281,9 +339,17 @@ typedef uint cl_mem_fence_flags;
 typedef enum memory_scope {
   memory_scope_work_item = __OPENCL_MEMORY_SCOPE_WORK_ITEM,
   memory_scope_work_group = __OPENCL_MEMORY_SCOPE_WORK_GROUP,
+#ifdef __opencl_c_atomic_scope_device
   memory_scope_device = __OPENCL_MEMORY_SCOPE_DEVICE,
+#endif
+#ifdef __opencl_c_atomic_scope_all_devices
+  #if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+  memory_scope_all_devices = __OPENCL_MEMORY_SCOPE_ALL_SVM_DEVICES,
+  #endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
   memory_scope_all_svm_devices = __OPENCL_MEMORY_SCOPE_ALL_SVM_DEVICES,
-#if defined(cl_intel_subgroups) || defined(cl_khr_subgroups)
+#endif
+#if defined(cl_intel_subgroups) || defined(cl_khr_subgroups) ||                \
+    defined(__opencl_c_subgroups)
   memory_scope_sub_group = __OPENCL_MEMORY_SCOPE_SUB_GROUP
 #endif
 } memory_scope;
@@ -301,13 +367,14 @@ typedef enum memory_scope {
 #define ATOMIC_FLAG_INIT 0
 
 // enum values aligned with what clang uses in EmitAtomicExpr()
-typedef enum memory_order
-{
+typedef enum memory_order {
   memory_order_relaxed = __ATOMIC_RELAXED,
   memory_order_acquire = __ATOMIC_ACQUIRE,
   memory_order_release = __ATOMIC_RELEASE,
   memory_order_acq_rel = __ATOMIC_ACQ_REL,
+#ifdef __opencl_c_atomic_order_seq_cst
   memory_order_seq_cst = __ATOMIC_SEQ_CST
+#endif //__opencl_c_atomic_order_seq_cst
 } memory_order;
 
 #endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
diff --git a/clang/lib/Headers/opencl-c.h b/clang/lib/Headers/opencl-c.h
index 3210f93cc851..93a946cec5b1 100644
--- a/clang/lib/Headers/opencl-c.h
+++ b/clang/lib/Headers/opencl-c.h
@@ -35,7 +35,6 @@
 #define __purefn __attribute__((pure))
 #define __cnfn __attribute__((const))
 
-
 // OpenCL v1.1/1.2/2.0 s6.2.3 - Explicit conversions
 
 char __ovld __cnfn convert_char_rte(char);
@@ -4632,7 +4631,7 @@ float16 __ovld __cnfn convert_float16(float16);
 
 // Conversions with double data type parameters or return value.
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 char __ovld __cnfn convert_char(double);
 char __ovld __cnfn convert_char_rte(double);
 char __ovld __cnfn convert_char_rtn(double);
@@ -5452,7 +5451,7 @@ double16 __ovld __cnfn convert_double16_rtz(uchar16);
 double16 __ovld __cnfn convert_double16_rtz(uint16);
 double16 __ovld __cnfn convert_double16_rtz(ulong16);
 double16 __ovld __cnfn convert_double16_rtz(ushort16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #ifdef cl_khr_fp16
 // Convert half types to non-double types.
@@ -6270,7 +6269,7 @@ half16 __ovld __cnfn convert_half16_rtz(float16);
 half16 __ovld __cnfn convert_half16_rtz(half16);
 
 // Convert half types to double types.
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn convert_double(half);
 double __ovld __cnfn convert_double_rte(half);
 double __ovld __cnfn convert_double_rtp(half);
@@ -6333,7 +6332,7 @@ half16 __ovld __cnfn convert_half16_rte(double16);
 half16 __ovld __cnfn convert_half16_rtp(double16);
 half16 __ovld __cnfn convert_half16_rtn(double16);
 half16 __ovld __cnfn convert_half16_rtz(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #endif // cl_khr_fp16
 
@@ -6404,14 +6403,14 @@ half16 __ovld __cnfn convert_half16_rtz(double16);
 #define as_float8(x) __builtin_astype((x),  float8)
 #define as_float16(x) __builtin_astype((x), float16)
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #define as_double(x) __builtin_astype((x),   double)
 #define as_double2(x) __builtin_astype((x),  double2)
 #define as_double3(x) __builtin_astype((x),  double3)
 #define as_double4(x) __builtin_astype((x),  double4)
 #define as_double8(x) __builtin_astype((x),  double8)
 #define as_double16(x) __builtin_astype((x), double16)
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #ifdef cl_khr_fp16
 #define as_half(x) __builtin_astype((x),   half)
@@ -6534,14 +6533,14 @@ float3 __ovld __cnfn acos(float3);
 float4 __ovld __cnfn acos(float4);
 float8 __ovld __cnfn acos(float8);
 float16 __ovld __cnfn acos(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn acos(double);
 double2 __ovld __cnfn acos(double2);
 double3 __ovld __cnfn acos(double3);
 double4 __ovld __cnfn acos(double4);
 double8 __ovld __cnfn acos(double8);
 double16 __ovld __cnfn acos(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn acos(half);
 half2 __ovld __cnfn acos(half2);
@@ -6560,14 +6559,14 @@ float3 __ovld __cnfn acosh(float3);
 float4 __ovld __cnfn acosh(float4);
 float8 __ovld __cnfn acosh(float8);
 float16 __ovld __cnfn acosh(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn acosh(double);
 double2 __ovld __cnfn acosh(double2);
 double3 __ovld __cnfn acosh(double3);
 double4 __ovld __cnfn acosh(double4);
 double8 __ovld __cnfn acosh(double8);
 double16 __ovld __cnfn acosh(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn acosh(half);
 half2 __ovld __cnfn acosh(half2);
@@ -6586,14 +6585,14 @@ float3 __ovld __cnfn acospi(float3 x);
 float4 __ovld __cnfn acospi(float4 x);
 float8 __ovld __cnfn acospi(float8 x);
 float16 __ovld __cnfn acospi(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn acospi(double x);
 double2 __ovld __cnfn acospi(double2 x);
 double3 __ovld __cnfn acospi(double3 x);
 double4 __ovld __cnfn acospi(double4 x);
 double8 __ovld __cnfn acospi(double8 x);
 double16 __ovld __cnfn acospi(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn acospi(half x);
 half2 __ovld __cnfn acospi(half2 x);
@@ -6612,14 +6611,14 @@ float3 __ovld __cnfn asin(float3);
 float4 __ovld __cnfn asin(float4);
 float8 __ovld __cnfn asin(float8);
 float16 __ovld __cnfn asin(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn asin(double);
 double2 __ovld __cnfn asin(double2);
 double3 __ovld __cnfn asin(double3);
 double4 __ovld __cnfn asin(double4);
 double8 __ovld __cnfn asin(double8);
 double16 __ovld __cnfn asin(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn asin(half);
 half2 __ovld __cnfn asin(half2);
@@ -6638,14 +6637,14 @@ float3 __ovld __cnfn asinh(float3);
 float4 __ovld __cnfn asinh(float4);
 float8 __ovld __cnfn asinh(float8);
 float16 __ovld __cnfn asinh(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn asinh(double);
 double2 __ovld __cnfn asinh(double2);
 double3 __ovld __cnfn asinh(double3);
 double4 __ovld __cnfn asinh(double4);
 double8 __ovld __cnfn asinh(double8);
 double16 __ovld __cnfn asinh(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn asinh(half);
 half2 __ovld __cnfn asinh(half2);
@@ -6664,14 +6663,14 @@ float3 __ovld __cnfn asinpi(float3 x);
 float4 __ovld __cnfn asinpi(float4 x);
 float8 __ovld __cnfn asinpi(float8 x);
 float16 __ovld __cnfn asinpi(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn asinpi(double x);
 double2 __ovld __cnfn asinpi(double2 x);
 double3 __ovld __cnfn asinpi(double3 x);
 double4 __ovld __cnfn asinpi(double4 x);
 double8 __ovld __cnfn asinpi(double8 x);
 double16 __ovld __cnfn asinpi(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn asinpi(half x);
 half2 __ovld __cnfn asinpi(half2 x);
@@ -6690,14 +6689,14 @@ float3 __ovld __cnfn atan(float3 y_over_x);
 float4 __ovld __cnfn atan(float4 y_over_x);
 float8 __ovld __cnfn atan(float8 y_over_x);
 float16 __ovld __cnfn atan(float16 y_over_x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn atan(double y_over_x);
 double2 __ovld __cnfn atan(double2 y_over_x);
 double3 __ovld __cnfn atan(double3 y_over_x);
 double4 __ovld __cnfn atan(double4 y_over_x);
 double8 __ovld __cnfn atan(double8 y_over_x);
 double16 __ovld __cnfn atan(double16 y_over_x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn atan(half y_over_x);
 half2 __ovld __cnfn atan(half2 y_over_x);
@@ -6716,14 +6715,14 @@ float3 __ovld __cnfn atan2(float3 y, float3 x);
 float4 __ovld __cnfn atan2(float4 y, float4 x);
 float8 __ovld __cnfn atan2(float8 y, float8 x);
 float16 __ovld __cnfn atan2(float16 y, float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn atan2(double y, double x);
 double2 __ovld __cnfn atan2(double2 y, double2 x);
 double3 __ovld __cnfn atan2(double3 y, double3 x);
 double4 __ovld __cnfn atan2(double4 y, double4 x);
 double8 __ovld __cnfn atan2(double8 y, double8 x);
 double16 __ovld __cnfn atan2(double16 y, double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn atan2(half y, half x);
 half2 __ovld __cnfn atan2(half2 y, half2 x);
@@ -6742,14 +6741,14 @@ float3 __ovld __cnfn atanh(float3);
 float4 __ovld __cnfn atanh(float4);
 float8 __ovld __cnfn atanh(float8);
 float16 __ovld __cnfn atanh(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn atanh(double);
 double2 __ovld __cnfn atanh(double2);
 double3 __ovld __cnfn atanh(double3);
 double4 __ovld __cnfn atanh(double4);
 double8 __ovld __cnfn atanh(double8);
 double16 __ovld __cnfn atanh(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn atanh(half);
 half2 __ovld __cnfn atanh(half2);
@@ -6768,14 +6767,14 @@ float3 __ovld __cnfn atanpi(float3 x);
 float4 __ovld __cnfn atanpi(float4 x);
 float8 __ovld __cnfn atanpi(float8 x);
 float16 __ovld __cnfn atanpi(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn atanpi(double x);
 double2 __ovld __cnfn atanpi(double2 x);
 double3 __ovld __cnfn atanpi(double3 x);
 double4 __ovld __cnfn atanpi(double4 x);
 double8 __ovld __cnfn atanpi(double8 x);
 double16 __ovld __cnfn atanpi(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn atanpi(half x);
 half2 __ovld __cnfn atanpi(half2 x);
@@ -6794,14 +6793,14 @@ float3 __ovld __cnfn atan2pi(float3 y, float3 x);
 float4 __ovld __cnfn atan2pi(float4 y, float4 x);
 float8 __ovld __cnfn atan2pi(float8 y, float8 x);
 float16 __ovld __cnfn atan2pi(float16 y, float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn atan2pi(double y, double x);
 double2 __ovld __cnfn atan2pi(double2 y, double2 x);
 double3 __ovld __cnfn atan2pi(double3 y, double3 x);
 double4 __ovld __cnfn atan2pi(double4 y, double4 x);
 double8 __ovld __cnfn atan2pi(double8 y, double8 x);
 double16 __ovld __cnfn atan2pi(double16 y, double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn atan2pi(half y, half x);
 half2 __ovld __cnfn atan2pi(half2 y, half2 x);
@@ -6820,14 +6819,14 @@ float3 __ovld __cnfn cbrt(float3);
 float4 __ovld __cnfn cbrt(float4);
 float8 __ovld __cnfn cbrt(float8);
 float16 __ovld __cnfn cbrt(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn cbrt(double);
 double2 __ovld __cnfn cbrt(double2);
 double3 __ovld __cnfn cbrt(double3);
 double4 __ovld __cnfn cbrt(double4);
 double8 __ovld __cnfn cbrt(double8);
 double16 __ovld __cnfn cbrt(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn cbrt(half);
 half2 __ovld __cnfn cbrt(half2);
@@ -6847,14 +6846,14 @@ float3 __ovld __cnfn ceil(float3);
 float4 __ovld __cnfn ceil(float4);
 float8 __ovld __cnfn ceil(float8);
 float16 __ovld __cnfn ceil(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn ceil(double);
 double2 __ovld __cnfn ceil(double2);
 double3 __ovld __cnfn ceil(double3);
 double4 __ovld __cnfn ceil(double4);
 double8 __ovld __cnfn ceil(double8);
 double16 __ovld __cnfn ceil(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn ceil(half);
 half2 __ovld __cnfn ceil(half2);
@@ -6873,14 +6872,14 @@ float3 __ovld __cnfn copysign(float3 x, float3 y);
 float4 __ovld __cnfn copysign(float4 x, float4 y);
 float8 __ovld __cnfn copysign(float8 x, float8 y);
 float16 __ovld __cnfn copysign(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn copysign(double x, double y);
 double2 __ovld __cnfn copysign(double2 x, double2 y);
 double3 __ovld __cnfn copysign(double3 x, double3 y);
 double4 __ovld __cnfn copysign(double4 x, double4 y);
 double8 __ovld __cnfn copysign(double8 x, double8 y);
 double16 __ovld __cnfn copysign(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn copysign(half x, half y);
 half2 __ovld __cnfn copysign(half2 x, half2 y);
@@ -6899,14 +6898,14 @@ float3 __ovld __cnfn cos(float3);
 float4 __ovld __cnfn cos(float4);
 float8 __ovld __cnfn cos(float8);
 float16 __ovld __cnfn cos(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn cos(double);
 double2 __ovld __cnfn cos(double2);
 double3 __ovld __cnfn cos(double3);
 double4 __ovld __cnfn cos(double4);
 double8 __ovld __cnfn cos(double8);
 double16 __ovld __cnfn cos(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn cos(half);
 half2 __ovld __cnfn cos(half2);
@@ -6925,14 +6924,14 @@ float3 __ovld __cnfn cosh(float3);
 float4 __ovld __cnfn cosh(float4);
 float8 __ovld __cnfn cosh(float8);
 float16 __ovld __cnfn cosh(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn cosh(double);
 double2 __ovld __cnfn cosh(double2);
 double3 __ovld __cnfn cosh(double3);
 double4 __ovld __cnfn cosh(double4);
 double8 __ovld __cnfn cosh(double8);
 double16 __ovld __cnfn cosh(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn cosh(half);
 half2 __ovld __cnfn cosh(half2);
@@ -6951,14 +6950,14 @@ float3 __ovld __cnfn cospi(float3 x);
 float4 __ovld __cnfn cospi(float4 x);
 float8 __ovld __cnfn cospi(float8 x);
 float16 __ovld __cnfn cospi(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn cospi(double x);
 double2 __ovld __cnfn cospi(double2 x);
 double3 __ovld __cnfn cospi(double3 x);
 double4 __ovld __cnfn cospi(double4 x);
 double8 __ovld __cnfn cospi(double8 x);
 double16 __ovld __cnfn cospi(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn cospi(half x);
 half2 __ovld __cnfn cospi(half2 x);
@@ -6977,14 +6976,14 @@ float3 __ovld __cnfn erfc(float3);
 float4 __ovld __cnfn erfc(float4);
 float8 __ovld __cnfn erfc(float8);
 float16 __ovld __cnfn erfc(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn erfc(double);
 double2 __ovld __cnfn erfc(double2);
 double3 __ovld __cnfn erfc(double3);
 double4 __ovld __cnfn erfc(double4);
 double8 __ovld __cnfn erfc(double8);
 double16 __ovld __cnfn erfc(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn erfc(half);
 half2 __ovld __cnfn erfc(half2);
@@ -7004,14 +7003,14 @@ float3 __ovld __cnfn erf(float3);
 float4 __ovld __cnfn erf(float4);
 float8 __ovld __cnfn erf(float8);
 float16 __ovld __cnfn erf(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn erf(double);
 double2 __ovld __cnfn erf(double2);
 double3 __ovld __cnfn erf(double3);
 double4 __ovld __cnfn erf(double4);
 double8 __ovld __cnfn erf(double8);
 double16 __ovld __cnfn erf(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn erf(half);
 half2 __ovld __cnfn erf(half2);
@@ -7030,14 +7029,14 @@ float3 __ovld __cnfn exp(float3 x);
 float4 __ovld __cnfn exp(float4 x);
 float8 __ovld __cnfn exp(float8 x);
 float16 __ovld __cnfn exp(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn exp(double x);
 double2 __ovld __cnfn exp(double2 x);
 double3 __ovld __cnfn exp(double3 x);
 double4 __ovld __cnfn exp(double4 x);
 double8 __ovld __cnfn exp(double8 x);
 double16 __ovld __cnfn exp(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn exp(half x);
 half2 __ovld __cnfn exp(half2 x);
@@ -7056,14 +7055,14 @@ float3 __ovld __cnfn exp2(float3);
 float4 __ovld __cnfn exp2(float4);
 float8 __ovld __cnfn exp2(float8);
 float16 __ovld __cnfn exp2(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn exp2(double);
 double2 __ovld __cnfn exp2(double2);
 double3 __ovld __cnfn exp2(double3);
 double4 __ovld __cnfn exp2(double4);
 double8 __ovld __cnfn exp2(double8);
 double16 __ovld __cnfn exp2(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn exp2(half);
 half2 __ovld __cnfn exp2(half2);
@@ -7082,14 +7081,14 @@ float3 __ovld __cnfn exp10(float3);
 float4 __ovld __cnfn exp10(float4);
 float8 __ovld __cnfn exp10(float8);
 float16 __ovld __cnfn exp10(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn exp10(double);
 double2 __ovld __cnfn exp10(double2);
 double3 __ovld __cnfn exp10(double3);
 double4 __ovld __cnfn exp10(double4);
 double8 __ovld __cnfn exp10(double8);
 double16 __ovld __cnfn exp10(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn exp10(half);
 half2 __ovld __cnfn exp10(half2);
@@ -7108,14 +7107,14 @@ float3 __ovld __cnfn expm1(float3 x);
 float4 __ovld __cnfn expm1(float4 x);
 float8 __ovld __cnfn expm1(float8 x);
 float16 __ovld __cnfn expm1(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn expm1(double x);
 double2 __ovld __cnfn expm1(double2 x);
 double3 __ovld __cnfn expm1(double3 x);
 double4 __ovld __cnfn expm1(double4 x);
 double8 __ovld __cnfn expm1(double8 x);
 double16 __ovld __cnfn expm1(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn expm1(half x);
 half2 __ovld __cnfn expm1(half2 x);
@@ -7134,14 +7133,14 @@ float3 __ovld __cnfn fabs(float3);
 float4 __ovld __cnfn fabs(float4);
 float8 __ovld __cnfn fabs(float8);
 float16 __ovld __cnfn fabs(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn fabs(double);
 double2 __ovld __cnfn fabs(double2);
 double3 __ovld __cnfn fabs(double3);
 double4 __ovld __cnfn fabs(double4);
 double8 __ovld __cnfn fabs(double8);
 double16 __ovld __cnfn fabs(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn fabs(half);
 half2 __ovld __cnfn fabs(half2);
@@ -7160,14 +7159,14 @@ float3 __ovld __cnfn fdim(float3 x, float3 y);
 float4 __ovld __cnfn fdim(float4 x, float4 y);
 float8 __ovld __cnfn fdim(float8 x, float8 y);
 float16 __ovld __cnfn fdim(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn fdim(double x, double y);
 double2 __ovld __cnfn fdim(double2 x, double2 y);
 double3 __ovld __cnfn fdim(double3 x, double3 y);
 double4 __ovld __cnfn fdim(double4 x, double4 y);
 double8 __ovld __cnfn fdim(double8 x, double8 y);
 double16 __ovld __cnfn fdim(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn fdim(half x, half y);
 half2 __ovld __cnfn fdim(half2 x, half2 y);
@@ -7187,14 +7186,14 @@ float3 __ovld __cnfn floor(float3);
 float4 __ovld __cnfn floor(float4);
 float8 __ovld __cnfn floor(float8);
 float16 __ovld __cnfn floor(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn floor(double);
 double2 __ovld __cnfn floor(double2);
 double3 __ovld __cnfn floor(double3);
 double4 __ovld __cnfn floor(double4);
 double8 __ovld __cnfn floor(double8);
 double16 __ovld __cnfn floor(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn floor(half);
 half2 __ovld __cnfn floor(half2);
@@ -7217,14 +7216,14 @@ float3 __ovld __cnfn fma(float3 a, float3 b, float3 c);
 float4 __ovld __cnfn fma(float4 a, float4 b, float4 c);
 float8 __ovld __cnfn fma(float8 a, float8 b, float8 c);
 float16 __ovld __cnfn fma(float16 a, float16 b, float16 c);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn fma(double a, double b, double c);
 double2 __ovld __cnfn fma(double2 a, double2 b, double2 c);
 double3 __ovld __cnfn fma(double3 a, double3 b, double3 c);
 double4 __ovld __cnfn fma(double4 a, double4 b, double4 c);
 double8 __ovld __cnfn fma(double8 a, double8 b, double8 c);
 double16 __ovld __cnfn fma(double16 a, double16 b, double16 c);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn fma(half a, half b, half c);
 half2 __ovld __cnfn fma(half2 a, half2 b, half2 c);
@@ -7251,7 +7250,7 @@ float3 __ovld __cnfn fmax(float3 x, float y);
 float4 __ovld __cnfn fmax(float4 x, float y);
 float8 __ovld __cnfn fmax(float8 x, float y);
 float16 __ovld __cnfn fmax(float16 x, float y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn fmax(double x, double y);
 double2 __ovld __cnfn fmax(double2 x, double2 y);
 double3 __ovld __cnfn fmax(double3 x, double3 y);
@@ -7263,7 +7262,7 @@ double3 __ovld __cnfn fmax(double3 x, double y);
 double4 __ovld __cnfn fmax(double4 x, double y);
 double8 __ovld __cnfn fmax(double8 x, double y);
 double16 __ovld __cnfn fmax(double16 x, double y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn fmax(half x, half y);
 half2 __ovld __cnfn fmax(half2 x, half2 y);
@@ -7295,7 +7294,7 @@ float3 __ovld __cnfn fmin(float3 x, float y);
 float4 __ovld __cnfn fmin(float4 x, float y);
 float8 __ovld __cnfn fmin(float8 x, float y);
 float16 __ovld __cnfn fmin(float16 x, float y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn fmin(double x, double y);
 double2 __ovld __cnfn fmin(double2 x, double2 y);
 double3 __ovld __cnfn fmin(double3 x, double3 y);
@@ -7307,7 +7306,7 @@ double3 __ovld __cnfn fmin(double3 x, double y);
 double4 __ovld __cnfn fmin(double4 x, double y);
 double8 __ovld __cnfn fmin(double8 x, double y);
 double16 __ovld __cnfn fmin(double16 x, double y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn fmin(half x, half y);
 half2 __ovld __cnfn fmin(half2 x, half2 y);
@@ -7331,14 +7330,14 @@ float3 __ovld __cnfn fmod(float3 x, float3 y);
 float4 __ovld __cnfn fmod(float4 x, float4 y);
 float8 __ovld __cnfn fmod(float8 x, float8 y);
 float16 __ovld __cnfn fmod(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn fmod(double x, double y);
 double2 __ovld __cnfn fmod(double2 x, double2 y);
 double3 __ovld __cnfn fmod(double3 x, double3 y);
 double4 __ovld __cnfn fmod(double4 x, double4 y);
 double8 __ovld __cnfn fmod(double8 x, double8 y);
 double16 __ovld __cnfn fmod(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn fmod(half x, half y);
 half2 __ovld __cnfn fmod(half2 x, half2 y);
@@ -7352,21 +7351,21 @@ half16 __ovld __cnfn fmod(half16 x, half16 y);
  * Returns fmin(x - floor (x), 0x1.fffffep-1f ).
  * floor(x) is returned in iptr.
  */
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float __ovld fract(float x, float *iptr);
 float2 __ovld fract(float2 x, float2 *iptr);
 float3 __ovld fract(float3 x, float3 *iptr);
 float4 __ovld fract(float4 x, float4 *iptr);
 float8 __ovld fract(float8 x, float8 *iptr);
 float16 __ovld fract(float16 x, float16 *iptr);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld fract(double x, double *iptr);
 double2 __ovld fract(double2 x, double2 *iptr);
 double3 __ovld fract(double3 x, double3 *iptr);
 double4 __ovld fract(double4 x, double4 *iptr);
 double8 __ovld fract(double8 x, double8 *iptr);
 double16 __ovld fract(double16 x, double16 *iptr);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld fract(half x, half *iptr);
 half2 __ovld fract(half2 x, half2 *iptr);
@@ -7375,7 +7374,9 @@ half4 __ovld fract(half4 x, half4 *iptr);
 half8 __ovld fract(half8 x, half8 *iptr);
 half16 __ovld fract(half16 x, half16 *iptr);
 #endif //cl_khr_fp16
-#else
+#endif //__opencl_c_generic_address_space
+
+#if !defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ != CL_VERSION_2_0)
 float __ovld fract(float x, __global float *iptr);
 float2 __ovld fract(float2 x, __global float2 *iptr);
 float3 __ovld fract(float3 x, __global float3 *iptr);
@@ -7394,7 +7395,7 @@ float3 __ovld fract(float3 x, __private float3 *iptr);
 float4 __ovld fract(float4 x, __private float4 *iptr);
 float8 __ovld fract(float8 x, __private float8 *iptr);
 float16 __ovld fract(float16 x, __private float16 *iptr);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld fract(double x, __global double *iptr);
 double2 __ovld fract(double2 x, __global double2 *iptr);
 double3 __ovld fract(double3 x, __global double3 *iptr);
@@ -7413,7 +7414,7 @@ double3 __ovld fract(double3 x, __private double3 *iptr);
 double4 __ovld fract(double4 x, __private double4 *iptr);
 double8 __ovld fract(double8 x, __private double8 *iptr);
 double16 __ovld fract(double16 x, __private double16 *iptr);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld fract(half x, __global half *iptr);
 half2 __ovld fract(half2 x, __global half2 *iptr);
@@ -7434,29 +7435,29 @@ half4 __ovld fract(half4 x, __private half4 *iptr);
 half8 __ovld fract(half8 x, __private half8 *iptr);
 half16 __ovld fract(half16 x, __private half16 *iptr);
 #endif //cl_khr_fp16
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
-
+#endif //! defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ !=
+       //! CL_VERSION_2_0)
 /**
  * Extract mantissa and exponent from x. For each
  * component the mantissa returned is a float with
  * magnitude in the interval [1/2, 1) or 0. Each
  * component of x equals mantissa returned * 2^exp.
  */
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float __ovld frexp(float x, int *exp);
 float2 __ovld frexp(float2 x, int2 *exp);
 float3 __ovld frexp(float3 x, int3 *exp);
 float4 __ovld frexp(float4 x, int4 *exp);
 float8 __ovld frexp(float8 x, int8 *exp);
 float16 __ovld frexp(float16 x, int16 *exp);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld frexp(double x, int *exp);
 double2 __ovld frexp(double2 x, int2 *exp);
 double3 __ovld frexp(double3 x, int3 *exp);
 double4 __ovld frexp(double4 x, int4 *exp);
 double8 __ovld frexp(double8 x, int8 *exp);
 double16 __ovld frexp(double16 x, int16 *exp);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld frexp(half x, int *exp);
 half2 __ovld frexp(half2 x, int2 *exp);
@@ -7465,7 +7466,9 @@ half4 __ovld frexp(half4 x, int4 *exp);
 half8 __ovld frexp(half8 x, int8 *exp);
 half16 __ovld frexp(half16 x, int16 *exp);
 #endif //cl_khr_fp16
-#else
+#endif //__opencl_c_generic_address_space
+
+#if !defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ != CL_VERSION_2_0)
 float __ovld frexp(float x, __global int *exp);
 float2 __ovld frexp(float2 x, __global int2 *exp);
 float3 __ovld frexp(float3 x, __global int3 *exp);
@@ -7484,7 +7487,7 @@ float3 __ovld frexp(float3 x, __private int3 *exp);
 float4 __ovld frexp(float4 x, __private int4 *exp);
 float8 __ovld frexp(float8 x, __private int8 *exp);
 float16 __ovld frexp(float16 x, __private int16 *exp);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld frexp(double x, __global int *exp);
 double2 __ovld frexp(double2 x, __global int2 *exp);
 double3 __ovld frexp(double3 x, __global int3 *exp);
@@ -7503,7 +7506,7 @@ double3 __ovld frexp(double3 x, __private int3 *exp);
 double4 __ovld frexp(double4 x, __private int4 *exp);
 double8 __ovld frexp(double8 x, __private int8 *exp);
 double16 __ovld frexp(double16 x, __private int16 *exp);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld frexp(half x, __global int *exp);
 half2 __ovld frexp(half2 x, __global int2 *exp);
@@ -7524,7 +7527,8 @@ half4 __ovld frexp(half4 x, __private int4 *exp);
 half8 __ovld frexp(half8 x, __private int8 *exp);
 half16 __ovld frexp(half16 x, __private int16 *exp);
 #endif //cl_khr_fp16
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#endif //! defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ !=
+       //! CL_VERSION_2_0)
 
 /**
  * Compute the value of the square root of x^2 + y^2
@@ -7536,14 +7540,14 @@ float3 __ovld __cnfn hypot(float3 x, float3 y);
 float4 __ovld __cnfn hypot(float4 x, float4 y);
 float8 __ovld __cnfn hypot(float8 x, float8 y);
 float16 __ovld __cnfn hypot(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn hypot(double x, double y);
 double2 __ovld __cnfn hypot(double2 x, double2 y);
 double3 __ovld __cnfn hypot(double3 x, double3 y);
 double4 __ovld __cnfn hypot(double4 x, double4 y);
 double8 __ovld __cnfn hypot(double8 x, double8 y);
 double16 __ovld __cnfn hypot(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn hypot(half x, half y);
 half2 __ovld __cnfn hypot(half2 x, half2 y);
@@ -7562,14 +7566,14 @@ int3 __ovld __cnfn ilogb(float3 x);
 int4 __ovld __cnfn ilogb(float4 x);
 int8 __ovld __cnfn ilogb(float8 x);
 int16 __ovld __cnfn ilogb(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn ilogb(double x);
 int2 __ovld __cnfn ilogb(double2 x);
 int3 __ovld __cnfn ilogb(double3 x);
 int4 __ovld __cnfn ilogb(double4 x);
 int8 __ovld __cnfn ilogb(double8 x);
 int16 __ovld __cnfn ilogb(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn ilogb(half x);
 int2 __ovld __cnfn ilogb(half2 x);
@@ -7593,7 +7597,7 @@ float3 __ovld __cnfn ldexp(float3 x, int n);
 float4 __ovld __cnfn ldexp(float4 x, int n);
 float8 __ovld __cnfn ldexp(float8 x, int n);
 float16 __ovld __cnfn ldexp(float16 x, int n);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn ldexp(double x, int n);
 double2 __ovld __cnfn ldexp(double2 x, int2 n);
 double3 __ovld __cnfn ldexp(double3 x, int3 n);
@@ -7605,7 +7609,7 @@ double3 __ovld __cnfn ldexp(double3 x, int n);
 double4 __ovld __cnfn ldexp(double4 x, int n);
 double8 __ovld __cnfn ldexp(double8 x, int n);
 double16 __ovld __cnfn ldexp(double16 x, int n);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn ldexp(half x, int n);
 half2 __ovld __cnfn ldexp(half2 x, int2 n);
@@ -7632,14 +7636,14 @@ float3 __ovld __cnfn lgamma(float3 x);
 float4 __ovld __cnfn lgamma(float4 x);
 float8 __ovld __cnfn lgamma(float8 x);
 float16 __ovld __cnfn lgamma(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn lgamma(double x);
 double2 __ovld __cnfn lgamma(double2 x);
 double3 __ovld __cnfn lgamma(double3 x);
 double4 __ovld __cnfn lgamma(double4 x);
 double8 __ovld __cnfn lgamma(double8 x);
 double16 __ovld __cnfn lgamma(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn lgamma(half x);
 half2 __ovld __cnfn lgamma(half2 x);
@@ -7649,21 +7653,21 @@ half8 __ovld __cnfn lgamma(half8 x);
 half16 __ovld __cnfn lgamma(half16 x);
 #endif //cl_khr_fp16
 
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float __ovld lgamma_r(float x, int *signp);
 float2 __ovld lgamma_r(float2 x, int2 *signp);
 float3 __ovld lgamma_r(float3 x, int3 *signp);
 float4 __ovld lgamma_r(float4 x, int4 *signp);
 float8 __ovld lgamma_r(float8 x, int8 *signp);
 float16 __ovld lgamma_r(float16 x, int16 *signp);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld lgamma_r(double x, int *signp);
 double2 __ovld lgamma_r(double2 x, int2 *signp);
 double3 __ovld lgamma_r(double3 x, int3 *signp);
 double4 __ovld lgamma_r(double4 x, int4 *signp);
 double8 __ovld lgamma_r(double8 x, int8 *signp);
 double16 __ovld lgamma_r(double16 x, int16 *signp);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld lgamma_r(half x, int *signp);
 half2 __ovld lgamma_r(half2 x, int2 *signp);
@@ -7672,7 +7676,9 @@ half4 __ovld lgamma_r(half4 x, int4 *signp);
 half8 __ovld lgamma_r(half8 x, int8 *signp);
 half16 __ovld lgamma_r(half16 x, int16 *signp);
 #endif //cl_khr_fp16
-#else
+#endif //__opencl_c_generic_address_space
+
+#if !defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ != CL_VERSION_2_0)
 float __ovld lgamma_r(float x, __global int *signp);
 float2 __ovld lgamma_r(float2 x, __global int2 *signp);
 float3 __ovld lgamma_r(float3 x, __global int3 *signp);
@@ -7691,7 +7697,7 @@ float3 __ovld lgamma_r(float3 x, __private int3 *signp);
 float4 __ovld lgamma_r(float4 x, __private int4 *signp);
 float8 __ovld lgamma_r(float8 x, __private int8 *signp);
 float16 __ovld lgamma_r(float16 x, __private int16 *signp);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld lgamma_r(double x, __global int *signp);
 double2 __ovld lgamma_r(double2 x, __global int2 *signp);
 double3 __ovld lgamma_r(double3 x, __global int3 *signp);
@@ -7710,7 +7716,7 @@ double3 __ovld lgamma_r(double3 x, __private int3 *signp);
 double4 __ovld lgamma_r(double4 x, __private int4 *signp);
 double8 __ovld lgamma_r(double8 x, __private int8 *signp);
 double16 __ovld lgamma_r(double16 x, __private int16 *signp);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld lgamma_r(half x, __global int *signp);
 half2 __ovld lgamma_r(half2 x, __global int2 *signp);
@@ -7731,8 +7737,8 @@ half4 __ovld lgamma_r(half4 x, __private int4 *signp);
 half8 __ovld lgamma_r(half8 x, __private int8 *signp);
 half16 __ovld lgamma_r(half16 x, __private int16 *signp);
 #endif //cl_khr_fp16
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
-
+#endif //! defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ !=
+       //! CL_VERSION_2_0)
 /**
  * Compute natural logarithm.
  */
@@ -7742,14 +7748,14 @@ float3 __ovld __cnfn log(float3);
 float4 __ovld __cnfn log(float4);
 float8 __ovld __cnfn log(float8);
 float16 __ovld __cnfn log(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn log(double);
 double2 __ovld __cnfn log(double2);
 double3 __ovld __cnfn log(double3);
 double4 __ovld __cnfn log(double4);
 double8 __ovld __cnfn log(double8);
 double16 __ovld __cnfn log(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn log(half);
 half2 __ovld __cnfn log(half2);
@@ -7760,7 +7766,7 @@ half16 __ovld __cnfn log(half16);
 #endif //cl_khr_fp16
 
 /**
- * Compute a base 2 logarithm.
+ * Compute a base 2 logarithm
  */
 float __ovld __cnfn log2(float);
 float2 __ovld __cnfn log2(float2);
@@ -7768,14 +7774,14 @@ float3 __ovld __cnfn log2(float3);
 float4 __ovld __cnfn log2(float4);
 float8 __ovld __cnfn log2(float8);
 float16 __ovld __cnfn log2(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn log2(double);
 double2 __ovld __cnfn log2(double2);
 double3 __ovld __cnfn log2(double3);
 double4 __ovld __cnfn log2(double4);
 double8 __ovld __cnfn log2(double8);
 double16 __ovld __cnfn log2(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn log2(half);
 half2 __ovld __cnfn log2(half2);
@@ -7794,14 +7800,14 @@ float3 __ovld __cnfn log10(float3);
 float4 __ovld __cnfn log10(float4);
 float8 __ovld __cnfn log10(float8);
 float16 __ovld __cnfn log10(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn log10(double);
 double2 __ovld __cnfn log10(double2);
 double3 __ovld __cnfn log10(double3);
 double4 __ovld __cnfn log10(double4);
 double8 __ovld __cnfn log10(double8);
 double16 __ovld __cnfn log10(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn log10(half);
 half2 __ovld __cnfn log10(half2);
@@ -7820,14 +7826,14 @@ float3 __ovld __cnfn log1p(float3 x);
 float4 __ovld __cnfn log1p(float4 x);
 float8 __ovld __cnfn log1p(float8 x);
 float16 __ovld __cnfn log1p(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn log1p(double x);
 double2 __ovld __cnfn log1p(double2 x);
 double3 __ovld __cnfn log1p(double3 x);
 double4 __ovld __cnfn log1p(double4 x);
 double8 __ovld __cnfn log1p(double8 x);
 double16 __ovld __cnfn log1p(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn log1p(half x);
 half2 __ovld __cnfn log1p(half2 x);
@@ -7847,14 +7853,14 @@ float3 __ovld __cnfn logb(float3 x);
 float4 __ovld __cnfn logb(float4 x);
 float8 __ovld __cnfn logb(float8 x);
 float16 __ovld __cnfn logb(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn logb(double x);
 double2 __ovld __cnfn logb(double2 x);
 double3 __ovld __cnfn logb(double3 x);
 double4 __ovld __cnfn logb(double4 x);
 double8 __ovld __cnfn logb(double8 x);
 double16 __ovld __cnfn logb(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn logb(half x);
 half2 __ovld __cnfn logb(half2 x);
@@ -7877,14 +7883,14 @@ float3 __ovld __cnfn mad(float3 a, float3 b, float3 c);
 float4 __ovld __cnfn mad(float4 a, float4 b, float4 c);
 float8 __ovld __cnfn mad(float8 a, float8 b, float8 c);
 float16 __ovld __cnfn mad(float16 a, float16 b, float16 c);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn mad(double a, double b, double c);
 double2 __ovld __cnfn mad(double2 a, double2 b, double2 c);
 double3 __ovld __cnfn mad(double3 a, double3 b, double3 c);
 double4 __ovld __cnfn mad(double4 a, double4 b, double4 c);
 double8 __ovld __cnfn mad(double8 a, double8 b, double8 c);
 double16 __ovld __cnfn mad(double16 a, double16 b, double16 c);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn mad(half a, half b, half c);
 half2 __ovld __cnfn mad(half2 a, half2 b, half2 c);
@@ -7904,14 +7910,14 @@ float3 __ovld __cnfn maxmag(float3 x, float3 y);
 float4 __ovld __cnfn maxmag(float4 x, float4 y);
 float8 __ovld __cnfn maxmag(float8 x, float8 y);
 float16 __ovld __cnfn maxmag(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn maxmag(double x, double y);
 double2 __ovld __cnfn maxmag(double2 x, double2 y);
 double3 __ovld __cnfn maxmag(double3 x, double3 y);
 double4 __ovld __cnfn maxmag(double4 x, double4 y);
 double8 __ovld __cnfn maxmag(double8 x, double8 y);
 double16 __ovld __cnfn maxmag(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn maxmag(half x, half y);
 half2 __ovld __cnfn maxmag(half2 x, half2 y);
@@ -7931,14 +7937,14 @@ float3 __ovld __cnfn minmag(float3 x, float3 y);
 float4 __ovld __cnfn minmag(float4 x, float4 y);
 float8 __ovld __cnfn minmag(float8 x, float8 y);
 float16 __ovld __cnfn minmag(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn minmag(double x, double y);
 double2 __ovld __cnfn minmag(double2 x, double2 y);
 double3 __ovld __cnfn minmag(double3 x, double3 y);
 double4 __ovld __cnfn minmag(double4 x, double4 y);
 double8 __ovld __cnfn minmag(double8 x, double8 y);
 double16 __ovld __cnfn minmag(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn minmag(half x, half y);
 half2 __ovld __cnfn minmag(half2 x, half2 y);
@@ -7955,21 +7961,21 @@ half16 __ovld __cnfn minmag(half16 x, half16 y);
  * the argument. It stores the integral part in the object
  * pointed to by iptr.
  */
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float __ovld modf(float x, float *iptr);
 float2 __ovld modf(float2 x, float2 *iptr);
 float3 __ovld modf(float3 x, float3 *iptr);
 float4 __ovld modf(float4 x, float4 *iptr);
 float8 __ovld modf(float8 x, float8 *iptr);
 float16 __ovld modf(float16 x, float16 *iptr);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld modf(double x, double *iptr);
 double2 __ovld modf(double2 x, double2 *iptr);
 double3 __ovld modf(double3 x, double3 *iptr);
 double4 __ovld modf(double4 x, double4 *iptr);
 double8 __ovld modf(double8 x, double8 *iptr);
 double16 __ovld modf(double16 x, double16 *iptr);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld modf(half x, half *iptr);
 half2 __ovld modf(half2 x, half2 *iptr);
@@ -7978,7 +7984,9 @@ half4 __ovld modf(half4 x, half4 *iptr);
 half8 __ovld modf(half8 x, half8 *iptr);
 half16 __ovld modf(half16 x, half16 *iptr);
 #endif //cl_khr_fp16
-#else
+#endif //__opencl_c_generic_address_space
+
+#if !defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ != CL_VERSION_2_0)
 float __ovld modf(float x, __global float *iptr);
 float2 __ovld modf(float2 x, __global float2 *iptr);
 float3 __ovld modf(float3 x, __global float3 *iptr);
@@ -7997,7 +8005,7 @@ float3 __ovld modf(float3 x, __private float3 *iptr);
 float4 __ovld modf(float4 x, __private float4 *iptr);
 float8 __ovld modf(float8 x, __private float8 *iptr);
 float16 __ovld modf(float16 x, __private float16 *iptr);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld modf(double x, __global double *iptr);
 double2 __ovld modf(double2 x, __global double2 *iptr);
 double3 __ovld modf(double3 x, __global double3 *iptr);
@@ -8016,7 +8024,7 @@ double3 __ovld modf(double3 x, __private double3 *iptr);
 double4 __ovld modf(double4 x, __private double4 *iptr);
 double8 __ovld modf(double8 x, __private double8 *iptr);
 double16 __ovld modf(double16 x, __private double16 *iptr);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld modf(half x, __global half *iptr);
 half2 __ovld modf(half2 x, __global half2 *iptr);
@@ -8037,7 +8045,8 @@ half4 __ovld modf(half4 x, __private half4 *iptr);
 half8 __ovld modf(half8 x, __private half8 *iptr);
 half16 __ovld modf(half16 x, __private half16 *iptr);
 #endif //cl_khr_fp16
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#endif //! defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ !=
+       //! CL_VERSION_2_0)
 
 /**
  * Returns a quiet NaN. The nancode may be placed
@@ -8049,14 +8058,14 @@ float3 __ovld __cnfn nan(uint3 nancode);
 float4 __ovld __cnfn nan(uint4 nancode);
 float8 __ovld __cnfn nan(uint8 nancode);
 float16 __ovld __cnfn nan(uint16 nancode);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn nan(ulong nancode);
 double2 __ovld __cnfn nan(ulong2 nancode);
 double3 __ovld __cnfn nan(ulong3 nancode);
 double4 __ovld __cnfn nan(ulong4 nancode);
 double8 __ovld __cnfn nan(ulong8 nancode);
 double16 __ovld __cnfn nan(ulong16 nancode);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn nan(ushort nancode);
 half2 __ovld __cnfn nan(ushort2 nancode);
@@ -8079,14 +8088,14 @@ float3 __ovld __cnfn nextafter(float3 x, float3 y);
 float4 __ovld __cnfn nextafter(float4 x, float4 y);
 float8 __ovld __cnfn nextafter(float8 x, float8 y);
 float16 __ovld __cnfn nextafter(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn nextafter(double x, double y);
 double2 __ovld __cnfn nextafter(double2 x, double2 y);
 double3 __ovld __cnfn nextafter(double3 x, double3 y);
 double4 __ovld __cnfn nextafter(double4 x, double4 y);
 double8 __ovld __cnfn nextafter(double8 x, double8 y);
 double16 __ovld __cnfn nextafter(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn nextafter(half x, half y);
 half2 __ovld __cnfn nextafter(half2 x, half2 y);
@@ -8105,14 +8114,14 @@ float3 __ovld __cnfn pow(float3 x, float3 y);
 float4 __ovld __cnfn pow(float4 x, float4 y);
 float8 __ovld __cnfn pow(float8 x, float8 y);
 float16 __ovld __cnfn pow(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn pow(double x, double y);
 double2 __ovld __cnfn pow(double2 x, double2 y);
 double3 __ovld __cnfn pow(double3 x, double3 y);
 double4 __ovld __cnfn pow(double4 x, double4 y);
 double8 __ovld __cnfn pow(double8 x, double8 y);
 double16 __ovld __cnfn pow(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn pow(half x, half y);
 half2 __ovld __cnfn pow(half2 x, half2 y);
@@ -8131,14 +8140,14 @@ float3 __ovld __cnfn pown(float3 x, int3 y);
 float4 __ovld __cnfn pown(float4 x, int4 y);
 float8 __ovld __cnfn pown(float8 x, int8 y);
 float16 __ovld __cnfn pown(float16 x, int16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn pown(double x, int y);
 double2 __ovld __cnfn pown(double2 x, int2 y);
 double3 __ovld __cnfn pown(double3 x, int3 y);
 double4 __ovld __cnfn pown(double4 x, int4 y);
 double8 __ovld __cnfn pown(double8 x, int8 y);
 double16 __ovld __cnfn pown(double16 x, int16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn pown(half x, int y);
 half2 __ovld __cnfn pown(half2 x, int2 y);
@@ -8157,14 +8166,14 @@ float3 __ovld __cnfn powr(float3 x, float3 y);
 float4 __ovld __cnfn powr(float4 x, float4 y);
 float8 __ovld __cnfn powr(float8 x, float8 y);
 float16 __ovld __cnfn powr(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn powr(double x, double y);
 double2 __ovld __cnfn powr(double2 x, double2 y);
 double3 __ovld __cnfn powr(double3 x, double3 y);
 double4 __ovld __cnfn powr(double4 x, double4 y);
 double8 __ovld __cnfn powr(double8 x, double8 y);
 double16 __ovld __cnfn powr(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn powr(half x, half y);
 half2 __ovld __cnfn powr(half2 x, half2 y);
@@ -8186,14 +8195,14 @@ float3 __ovld __cnfn remainder(float3 x, float3 y);
 float4 __ovld __cnfn remainder(float4 x, float4 y);
 float8 __ovld __cnfn remainder(float8 x, float8 y);
 float16 __ovld __cnfn remainder(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn remainder(double x, double y);
 double2 __ovld __cnfn remainder(double2 x, double2 y);
 double3 __ovld __cnfn remainder(double3 x, double3 y);
 double4 __ovld __cnfn remainder(double4 x, double4 y);
 double8 __ovld __cnfn remainder(double8 x, double8 y);
 double16 __ovld __cnfn remainder(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn remainder(half x, half y);
 half2 __ovld __cnfn remainder(half2 x, half2 y);
@@ -8215,21 +8224,21 @@ half16 __ovld __cnfn remainder(half16 x, half16 y);
  * sign as x/y. It stores this signed value in the object
  * pointed to by quo.
  */
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float __ovld remquo(float x, float y, int *quo);
 float2 __ovld remquo(float2 x, float2 y, int2 *quo);
 float3 __ovld remquo(float3 x, float3 y, int3 *quo);
 float4 __ovld remquo(float4 x, float4 y, int4 *quo);
 float8 __ovld remquo(float8 x, float8 y, int8 *quo);
 float16 __ovld remquo(float16 x, float16 y, int16 *quo);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld remquo(double x, double y, int *quo);
 double2 __ovld remquo(double2 x, double2 y, int2 *quo);
 double3 __ovld remquo(double3 x, double3 y, int3 *quo);
 double4 __ovld remquo(double4 x, double4 y, int4 *quo);
 double8 __ovld remquo(double8 x, double8 y, int8 *quo);
 double16 __ovld remquo(double16 x, double16 y, int16 *quo);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld remquo(half x, half y, int *quo);
 half2 __ovld remquo(half2 x, half2 y, int2 *quo);
@@ -8237,9 +8246,10 @@ half3 __ovld remquo(half3 x, half3 y, int3 *quo);
 half4 __ovld remquo(half4 x, half4 y, int4 *quo);
 half8 __ovld remquo(half8 x, half8 y, int8 *quo);
 half16 __ovld remquo(half16 x, half16 y, int16 *quo);
-
 #endif //cl_khr_fp16
-#else
+#endif //__opencl_c_generic_address_space
+
+#if !defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ != CL_VERSION_2_0)
 float __ovld remquo(float x, float y, __global int *quo);
 float2 __ovld remquo(float2 x, float2 y, __global int2 *quo);
 float3 __ovld remquo(float3 x, float3 y, __global int3 *quo);
@@ -8258,7 +8268,7 @@ float3 __ovld remquo(float3 x, float3 y, __private int3 *quo);
 float4 __ovld remquo(float4 x, float4 y, __private int4 *quo);
 float8 __ovld remquo(float8 x, float8 y, __private int8 *quo);
 float16 __ovld remquo(float16 x, float16 y, __private int16 *quo);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld remquo(double x, double y, __global int *quo);
 double2 __ovld remquo(double2 x, double2 y, __global int2 *quo);
 double3 __ovld remquo(double3 x, double3 y, __global int3 *quo);
@@ -8277,7 +8287,7 @@ double3 __ovld remquo(double3 x, double3 y, __private int3 *quo);
 double4 __ovld remquo(double4 x, double4 y, __private int4 *quo);
 double8 __ovld remquo(double8 x, double8 y, __private int8 *quo);
 double16 __ovld remquo(double16 x, double16 y, __private int16 *quo);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld remquo(half x, half y, __global int *quo);
 half2 __ovld remquo(half2 x, half2 y, __global int2 *quo);
@@ -8298,7 +8308,8 @@ half4 __ovld remquo(half4 x, half4 y, __private int4 *quo);
 half8 __ovld remquo(half8 x, half8 y, __private int8 *quo);
 half16 __ovld remquo(half16 x, half16 y, __private int16 *quo);
 #endif //cl_khr_fp16
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#endif //! defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ !=
+       //! CL_VERSION_2_0)
 /**
  * Round to integral value (using round to nearest
  * even rounding mode) in floating-point format.
@@ -8311,14 +8322,14 @@ float3 __ovld __cnfn rint(float3);
 float4 __ovld __cnfn rint(float4);
 float8 __ovld __cnfn rint(float8);
 float16 __ovld __cnfn rint(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn rint(double);
 double2 __ovld __cnfn rint(double2);
 double3 __ovld __cnfn rint(double3);
 double4 __ovld __cnfn rint(double4);
 double8 __ovld __cnfn rint(double8);
 double16 __ovld __cnfn rint(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn rint(half);
 half2 __ovld __cnfn rint(half2);
@@ -8337,14 +8348,14 @@ float3 __ovld __cnfn rootn(float3 x, int3 y);
 float4 __ovld __cnfn rootn(float4 x, int4 y);
 float8 __ovld __cnfn rootn(float8 x, int8 y);
 float16 __ovld __cnfn rootn(float16 x, int16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn rootn(double x, int y);
 double2 __ovld __cnfn rootn(double2 x, int2 y);
 double3 __ovld __cnfn rootn(double3 x, int3 y);
 double4 __ovld __cnfn rootn(double4 x, int4 y);
 double8 __ovld __cnfn rootn(double8 x, int8 y);
 double16 __ovld __cnfn rootn(double16 x, int16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn rootn(half x, int y);
 half2 __ovld __cnfn rootn(half2 x, int2 y);
@@ -8365,14 +8376,14 @@ float3 __ovld __cnfn round(float3 x);
 float4 __ovld __cnfn round(float4 x);
 float8 __ovld __cnfn round(float8 x);
 float16 __ovld __cnfn round(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn round(double x);
 double2 __ovld __cnfn round(double2 x);
 double3 __ovld __cnfn round(double3 x);
 double4 __ovld __cnfn round(double4 x);
 double8 __ovld __cnfn round(double8 x);
 double16 __ovld __cnfn round(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn round(half x);
 half2 __ovld __cnfn round(half2 x);
@@ -8391,14 +8402,14 @@ float3 __ovld __cnfn rsqrt(float3);
 float4 __ovld __cnfn rsqrt(float4);
 float8 __ovld __cnfn rsqrt(float8);
 float16 __ovld __cnfn rsqrt(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn rsqrt(double);
 double2 __ovld __cnfn rsqrt(double2);
 double3 __ovld __cnfn rsqrt(double3);
 double4 __ovld __cnfn rsqrt(double4);
 double8 __ovld __cnfn rsqrt(double8);
 double16 __ovld __cnfn rsqrt(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn rsqrt(half);
 half2 __ovld __cnfn rsqrt(half2);
@@ -8417,14 +8428,14 @@ float3 __ovld __cnfn sin(float3);
 float4 __ovld __cnfn sin(float4);
 float8 __ovld __cnfn sin(float8);
 float16 __ovld __cnfn sin(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn sin(double);
 double2 __ovld __cnfn sin(double2);
 double3 __ovld __cnfn sin(double3);
 double4 __ovld __cnfn sin(double4);
 double8 __ovld __cnfn sin(double8);
 double16 __ovld __cnfn sin(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn sin(half);
 half2 __ovld __cnfn sin(half2);
@@ -8439,21 +8450,21 @@ half16 __ovld __cnfn sin(half16);
  * is the return value and computed cosine is returned
  * in cosval.
  */
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float __ovld sincos(float x, float *cosval);
 float2 __ovld sincos(float2 x, float2 *cosval);
 float3 __ovld sincos(float3 x, float3 *cosval);
 float4 __ovld sincos(float4 x, float4 *cosval);
 float8 __ovld sincos(float8 x, float8 *cosval);
 float16 __ovld sincos(float16 x, float16 *cosval);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld sincos(double x, double *cosval);
 double2 __ovld sincos(double2 x, double2 *cosval);
 double3 __ovld sincos(double3 x, double3 *cosval);
 double4 __ovld sincos(double4 x, double4 *cosval);
 double8 __ovld sincos(double8 x, double8 *cosval);
 double16 __ovld sincos(double16 x, double16 *cosval);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld sincos(half x, half *cosval);
 half2 __ovld sincos(half2 x, half2 *cosval);
@@ -8462,7 +8473,9 @@ half4 __ovld sincos(half4 x, half4 *cosval);
 half8 __ovld sincos(half8 x, half8 *cosval);
 half16 __ovld sincos(half16 x, half16 *cosval);
 #endif //cl_khr_fp16
-#else
+#endif //__opencl_c_generic_address_space
+
+#if !defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ != CL_VERSION_2_0)
 float __ovld sincos(float x, __global float *cosval);
 float2 __ovld sincos(float2 x, __global float2 *cosval);
 float3 __ovld sincos(float3 x, __global float3 *cosval);
@@ -8481,7 +8494,7 @@ float3 __ovld sincos(float3 x, __private float3 *cosval);
 float4 __ovld sincos(float4 x, __private float4 *cosval);
 float8 __ovld sincos(float8 x, __private float8 *cosval);
 float16 __ovld sincos(float16 x, __private float16 *cosval);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld sincos(double x, __global double *cosval);
 double2 __ovld sincos(double2 x, __global double2 *cosval);
 double3 __ovld sincos(double3 x, __global double3 *cosval);
@@ -8500,7 +8513,7 @@ double3 __ovld sincos(double3 x, __private double3 *cosval);
 double4 __ovld sincos(double4 x, __private double4 *cosval);
 double8 __ovld sincos(double8 x, __private double8 *cosval);
 double16 __ovld sincos(double16 x, __private double16 *cosval);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld sincos(half x, __global half *cosval);
 half2 __ovld sincos(half2 x, __global half2 *cosval);
@@ -8521,8 +8534,8 @@ half4 __ovld sincos(half4 x, __private half4 *cosval);
 half8 __ovld sincos(half8 x, __private half8 *cosval);
 half16 __ovld sincos(half16 x, __private half16 *cosval);
 #endif //cl_khr_fp16
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
-
+#endif //! defined(__OPENCL_CPP_VERSION__) && (__OPENCL_C_VERSION__ !=
+       //! CL_VERSION_2_0)
 /**
  * Compute hyperbolic sine.
  */
@@ -8532,14 +8545,14 @@ float3 __ovld __cnfn sinh(float3);
 float4 __ovld __cnfn sinh(float4);
 float8 __ovld __cnfn sinh(float8);
 float16 __ovld __cnfn sinh(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn sinh(double);
 double2 __ovld __cnfn sinh(double2);
 double3 __ovld __cnfn sinh(double3);
 double4 __ovld __cnfn sinh(double4);
 double8 __ovld __cnfn sinh(double8);
 double16 __ovld __cnfn sinh(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn sinh(half);
 half2 __ovld __cnfn sinh(half2);
@@ -8558,14 +8571,14 @@ float3 __ovld __cnfn sinpi(float3 x);
 float4 __ovld __cnfn sinpi(float4 x);
 float8 __ovld __cnfn sinpi(float8 x);
 float16 __ovld __cnfn sinpi(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn sinpi(double x);
 double2 __ovld __cnfn sinpi(double2 x);
 double3 __ovld __cnfn sinpi(double3 x);
 double4 __ovld __cnfn sinpi(double4 x);
 double8 __ovld __cnfn sinpi(double8 x);
 double16 __ovld __cnfn sinpi(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn sinpi(half x);
 half2 __ovld __cnfn sinpi(half2 x);
@@ -8584,14 +8597,14 @@ float3 __ovld __cnfn sqrt(float3);
 float4 __ovld __cnfn sqrt(float4);
 float8 __ovld __cnfn sqrt(float8);
 float16 __ovld __cnfn sqrt(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn sqrt(double);
 double2 __ovld __cnfn sqrt(double2);
 double3 __ovld __cnfn sqrt(double3);
 double4 __ovld __cnfn sqrt(double4);
 double8 __ovld __cnfn sqrt(double8);
 double16 __ovld __cnfn sqrt(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn sqrt(half);
 half2 __ovld __cnfn sqrt(half2);
@@ -8610,14 +8623,14 @@ float3 __ovld __cnfn tan(float3);
 float4 __ovld __cnfn tan(float4);
 float8 __ovld __cnfn tan(float8);
 float16 __ovld __cnfn tan(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn tan(double);
 double2 __ovld __cnfn tan(double2);
 double3 __ovld __cnfn tan(double3);
 double4 __ovld __cnfn tan(double4);
 double8 __ovld __cnfn tan(double8);
 double16 __ovld __cnfn tan(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn tan(half);
 half2 __ovld __cnfn tan(half2);
@@ -8636,14 +8649,14 @@ float3 __ovld __cnfn tanh(float3);
 float4 __ovld __cnfn tanh(float4);
 float8 __ovld __cnfn tanh(float8);
 float16 __ovld __cnfn tanh(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn tanh(double);
 double2 __ovld __cnfn tanh(double2);
 double3 __ovld __cnfn tanh(double3);
 double4 __ovld __cnfn tanh(double4);
 double8 __ovld __cnfn tanh(double8);
 double16 __ovld __cnfn tanh(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn tanh(half);
 half2 __ovld __cnfn tanh(half2);
@@ -8662,14 +8675,14 @@ float3 __ovld __cnfn tanpi(float3 x);
 float4 __ovld __cnfn tanpi(float4 x);
 float8 __ovld __cnfn tanpi(float8 x);
 float16 __ovld __cnfn tanpi(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn tanpi(double x);
 double2 __ovld __cnfn tanpi(double2 x);
 double3 __ovld __cnfn tanpi(double3 x);
 double4 __ovld __cnfn tanpi(double4 x);
 double8 __ovld __cnfn tanpi(double8 x);
 double16 __ovld __cnfn tanpi(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn tanpi(half x);
 half2 __ovld __cnfn tanpi(half2 x);
@@ -8688,14 +8701,14 @@ float3 __ovld __cnfn tgamma(float3);
 float4 __ovld __cnfn tgamma(float4);
 float8 __ovld __cnfn tgamma(float8);
 float16 __ovld __cnfn tgamma(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn tgamma(double);
 double2 __ovld __cnfn tgamma(double2);
 double3 __ovld __cnfn tgamma(double3);
 double4 __ovld __cnfn tgamma(double4);
 double8 __ovld __cnfn tgamma(double8);
 double16 __ovld __cnfn tgamma(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn tgamma(half);
 half2 __ovld __cnfn tgamma(half2);
@@ -8715,14 +8728,14 @@ float3 __ovld __cnfn trunc(float3);
 float4 __ovld __cnfn trunc(float4);
 float8 __ovld __cnfn trunc(float8);
 float16 __ovld __cnfn trunc(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn trunc(double);
 double2 __ovld __cnfn trunc(double2);
 double3 __ovld __cnfn trunc(double3);
 double4 __ovld __cnfn trunc(double4);
 double8 __ovld __cnfn trunc(double8);
 double16 __ovld __cnfn trunc(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn trunc(half);
 half2 __ovld __cnfn trunc(half2);
@@ -10108,7 +10121,7 @@ float3 __ovld __cnfn clamp(float3 x, float minval, float maxval);
 float4 __ovld __cnfn clamp(float4 x, float minval, float maxval);
 float8 __ovld __cnfn clamp(float8 x, float minval, float maxval);
 float16 __ovld __cnfn clamp(float16 x, float minval, float maxval);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn clamp(double x, double minval, double maxval);
 double2 __ovld __cnfn clamp(double2 x, double2 minval, double2 maxval);
 double3 __ovld __cnfn clamp(double3 x, double3 minval, double3 maxval);
@@ -10120,7 +10133,7 @@ double3 __ovld __cnfn clamp(double3 x, double minval, double maxval);
 double4 __ovld __cnfn clamp(double4 x, double minval, double maxval);
 double8 __ovld __cnfn clamp(double8 x, double minval, double maxval);
 double16 __ovld __cnfn clamp(double16 x, double minval, double maxval);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn clamp(half x, half minval, half maxval);
 half2 __ovld __cnfn clamp(half2 x, half2 minval, half2 maxval);
@@ -10145,14 +10158,14 @@ float3 __ovld __cnfn degrees(float3 radians);
 float4 __ovld __cnfn degrees(float4 radians);
 float8 __ovld __cnfn degrees(float8 radians);
 float16 __ovld __cnfn degrees(float16 radians);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn degrees(double radians);
 double2 __ovld __cnfn degrees(double2 radians);
 double3 __ovld __cnfn degrees(double3 radians);
 double4 __ovld __cnfn degrees(double4 radians);
 double8 __ovld __cnfn degrees(double8 radians);
 double16 __ovld __cnfn degrees(double16 radians);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn degrees(half radians);
 half2 __ovld __cnfn degrees(half2 radians);
@@ -10177,7 +10190,7 @@ float3 __ovld __cnfn max(float3 x, float y);
 float4 __ovld __cnfn max(float4 x, float y);
 float8 __ovld __cnfn max(float8 x, float y);
 float16 __ovld __cnfn max(float16 x, float y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn max(double x, double y);
 double2 __ovld __cnfn max(double2 x, double2 y);
 double3 __ovld __cnfn max(double3 x, double3 y);
@@ -10189,7 +10202,7 @@ double3 __ovld __cnfn max(double3 x, double y);
 double4 __ovld __cnfn max(double4 x, double y);
 double8 __ovld __cnfn max(double8 x, double y);
 double16 __ovld __cnfn max(double16 x, double y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn max(half x, half y);
 half2 __ovld __cnfn max(half2 x, half2 y);
@@ -10219,7 +10232,7 @@ float3 __ovld __cnfn min(float3 x, float y);
 float4 __ovld __cnfn min(float4 x, float y);
 float8 __ovld __cnfn min(float8 x, float y);
 float16 __ovld __cnfn min(float16 x, float y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn min(double x, double y);
 double2 __ovld __cnfn min(double2 x, double2 y);
 double3 __ovld __cnfn min(double3 x, double3 y);
@@ -10231,7 +10244,7 @@ double3 __ovld __cnfn min(double3 x, double y);
 double4 __ovld __cnfn min(double4 x, double y);
 double8 __ovld __cnfn min(double8 x, double y);
 double16 __ovld __cnfn min(double16 x, double y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn min(half x, half y);
 half2 __ovld __cnfn min(half2 x, half2 y);
@@ -10264,7 +10277,7 @@ float3 __ovld __cnfn mix(float3 x, float3 y, float a);
 float4 __ovld __cnfn mix(float4 x, float4 y, float a);
 float8 __ovld __cnfn mix(float8 x, float8 y, float a);
 float16 __ovld __cnfn mix(float16 x, float16 y, float a);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn mix(double x, double y, double a);
 double2 __ovld __cnfn mix(double2 x, double2 y, double2 a);
 double3 __ovld __cnfn mix(double3 x, double3 y, double3 a);
@@ -10276,7 +10289,7 @@ double3 __ovld __cnfn mix(double3 x, double3 y, double a);
 double4 __ovld __cnfn mix(double4 x, double4 y, double a);
 double8 __ovld __cnfn mix(double8 x, double8 y, double a);
 double16 __ovld __cnfn mix(double16 x, double16 y, double a);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn mix(half x, half y, half a);
 half2 __ovld __cnfn mix(half2 x, half2 y, half2 a);
@@ -10301,14 +10314,14 @@ float3 __ovld __cnfn radians(float3 degrees);
 float4 __ovld __cnfn radians(float4 degrees);
 float8 __ovld __cnfn radians(float8 degrees);
 float16 __ovld __cnfn radians(float16 degrees);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn radians(double degrees);
 double2 __ovld __cnfn radians(double2 degrees);
 double3 __ovld __cnfn radians(double3 degrees);
 double4 __ovld __cnfn radians(double4 degrees);
 double8 __ovld __cnfn radians(double8 degrees);
 double16 __ovld __cnfn radians(double16 degrees);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn radians(half degrees);
 half2 __ovld __cnfn radians(half2 degrees);
@@ -10332,7 +10345,7 @@ float3 __ovld __cnfn step(float edge, float3 x);
 float4 __ovld __cnfn step(float edge, float4 x);
 float8 __ovld __cnfn step(float edge, float8 x);
 float16 __ovld __cnfn step(float edge, float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn step(double edge, double x);
 double2 __ovld __cnfn step(double2 edge, double2 x);
 double3 __ovld __cnfn step(double3 edge, double3 x);
@@ -10344,7 +10357,7 @@ double3 __ovld __cnfn step(double edge, double3 x);
 double4 __ovld __cnfn step(double edge, double4 x);
 double8 __ovld __cnfn step(double edge, double8 x);
 double16 __ovld __cnfn step(double edge, double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn step(half edge, half x);
 half2 __ovld __cnfn step(half2 edge, half2 x);
@@ -10383,7 +10396,7 @@ float3 __ovld __cnfn smoothstep(float edge0, float edge1, float3 x);
 float4 __ovld __cnfn smoothstep(float edge0, float edge1, float4 x);
 float8 __ovld __cnfn smoothstep(float edge0, float edge1, float8 x);
 float16 __ovld __cnfn smoothstep(float edge0, float edge1, float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn smoothstep(double edge0, double edge1, double x);
 double2 __ovld __cnfn smoothstep(double2 edge0, double2 edge1, double2 x);
 double3 __ovld __cnfn smoothstep(double3 edge0, double3 edge1, double3 x);
@@ -10395,7 +10408,7 @@ double3 __ovld __cnfn smoothstep(double edge0, double edge1, double3 x);
 double4 __ovld __cnfn smoothstep(double edge0, double edge1, double4 x);
 double8 __ovld __cnfn smoothstep(double edge0, double edge1, double8 x);
 double16 __ovld __cnfn smoothstep(double edge0, double edge1, double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn smoothstep(half edge0, half edge1, half x);
 half2 __ovld __cnfn smoothstep(half2 edge0, half2 edge1, half2 x);
@@ -10420,14 +10433,14 @@ float3 __ovld __cnfn sign(float3 x);
 float4 __ovld __cnfn sign(float4 x);
 float8 __ovld __cnfn sign(float8 x);
 float16 __ovld __cnfn sign(float16 x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn sign(double x);
 double2 __ovld __cnfn sign(double2 x);
 double3 __ovld __cnfn sign(double3 x);
 double4 __ovld __cnfn sign(double4 x);
 double8 __ovld __cnfn sign(double8 x);
 double16 __ovld __cnfn sign(double16 x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn sign(half x);
 half2 __ovld __cnfn sign(half2 x);
@@ -10445,10 +10458,10 @@ half16 __ovld __cnfn sign(half16 x);
  */
 float4 __ovld __cnfn cross(float4 p0, float4 p1);
 float3 __ovld __cnfn cross(float3 p0, float3 p1);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double4 __ovld __cnfn cross(double4 p0, double4 p1);
 double3 __ovld __cnfn cross(double3 p0, double3 p1);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half4 __ovld __cnfn cross(half4 p0, half4 p1);
 half3 __ovld __cnfn cross(half3 p0, half3 p1);
@@ -10461,12 +10474,12 @@ float __ovld __cnfn dot(float p0, float p1);
 float __ovld __cnfn dot(float2 p0, float2 p1);
 float __ovld __cnfn dot(float3 p0, float3 p1);
 float __ovld __cnfn dot(float4 p0, float4 p1);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn dot(double p0, double p1);
 double __ovld __cnfn dot(double2 p0, double2 p1);
 double __ovld __cnfn dot(double3 p0, double3 p1);
 double __ovld __cnfn dot(double4 p0, double4 p1);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn dot(half p0, half p1);
 half __ovld __cnfn dot(half2 p0, half2 p1);
@@ -10482,12 +10495,12 @@ float __ovld __cnfn distance(float p0, float p1);
 float __ovld __cnfn distance(float2 p0, float2 p1);
 float __ovld __cnfn distance(float3 p0, float3 p1);
 float __ovld __cnfn distance(float4 p0, float4 p1);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn distance(double p0, double p1);
 double __ovld __cnfn distance(double2 p0, double2 p1);
 double __ovld __cnfn distance(double3 p0, double3 p1);
 double __ovld __cnfn distance(double4 p0, double4 p1);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn distance(half p0, half p1);
 half __ovld __cnfn distance(half2 p0, half2 p1);
@@ -10503,12 +10516,12 @@ float __ovld __cnfn length(float p);
 float __ovld __cnfn length(float2 p);
 float __ovld __cnfn length(float3 p);
 float __ovld __cnfn length(float4 p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn length(double p);
 double __ovld __cnfn length(double2 p);
 double __ovld __cnfn length(double3 p);
 double __ovld __cnfn length(double4 p);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn length(half p);
 half __ovld __cnfn length(half2 p);
@@ -10524,12 +10537,12 @@ float __ovld __cnfn normalize(float p);
 float2 __ovld __cnfn normalize(float2 p);
 float3 __ovld __cnfn normalize(float3 p);
 float4 __ovld __cnfn normalize(float4 p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn normalize(double p);
 double2 __ovld __cnfn normalize(double2 p);
 double3 __ovld __cnfn normalize(double3 p);
 double4 __ovld __cnfn normalize(double4 p);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn normalize(half p);
 half2 __ovld __cnfn normalize(half2 p);
@@ -10610,14 +10623,14 @@ int3 __ovld __cnfn isequal(float3 x, float3 y);
 int4 __ovld __cnfn isequal(float4 x, float4 y);
 int8 __ovld __cnfn isequal(float8 x, float8 y);
 int16 __ovld __cnfn isequal(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isequal(double x, double y);
 long2 __ovld __cnfn isequal(double2 x, double2 y);
 long3 __ovld __cnfn isequal(double3 x, double3 y);
 long4 __ovld __cnfn isequal(double4 x, double4 y);
 long8 __ovld __cnfn isequal(double8 x, double8 y);
 long16 __ovld __cnfn isequal(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isequal(half x, half y);
 short2 __ovld __cnfn isequal(half2 x, half2 y);
@@ -10636,14 +10649,14 @@ int3 __ovld __cnfn isnotequal(float3 x, float3 y);
 int4 __ovld __cnfn isnotequal(float4 x, float4 y);
 int8 __ovld __cnfn isnotequal(float8 x, float8 y);
 int16 __ovld __cnfn isnotequal(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isnotequal(double x, double y);
 long2 __ovld __cnfn isnotequal(double2 x, double2 y);
 long3 __ovld __cnfn isnotequal(double3 x, double3 y);
 long4 __ovld __cnfn isnotequal(double4 x, double4 y);
 long8 __ovld __cnfn isnotequal(double8 x, double8 y);
 long16 __ovld __cnfn isnotequal(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isnotequal(half x, half y);
 short2 __ovld __cnfn isnotequal(half2 x, half2 y);
@@ -10662,14 +10675,14 @@ int3 __ovld __cnfn isgreater(float3 x, float3 y);
 int4 __ovld __cnfn isgreater(float4 x, float4 y);
 int8 __ovld __cnfn isgreater(float8 x, float8 y);
 int16 __ovld __cnfn isgreater(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isgreater(double x, double y);
 long2 __ovld __cnfn isgreater(double2 x, double2 y);
 long3 __ovld __cnfn isgreater(double3 x, double3 y);
 long4 __ovld __cnfn isgreater(double4 x, double4 y);
 long8 __ovld __cnfn isgreater(double8 x, double8 y);
 long16 __ovld __cnfn isgreater(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isgreater(half x, half y);
 short2 __ovld __cnfn isgreater(half2 x, half2 y);
@@ -10688,14 +10701,14 @@ int3 __ovld __cnfn isgreaterequal(float3 x, float3 y);
 int4 __ovld __cnfn isgreaterequal(float4 x, float4 y);
 int8 __ovld __cnfn isgreaterequal(float8 x, float8 y);
 int16 __ovld __cnfn isgreaterequal(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isgreaterequal(double x, double y);
 long2 __ovld __cnfn isgreaterequal(double2 x, double2 y);
 long3 __ovld __cnfn isgreaterequal(double3 x, double3 y);
 long4 __ovld __cnfn isgreaterequal(double4 x, double4 y);
 long8 __ovld __cnfn isgreaterequal(double8 x, double8 y);
 long16 __ovld __cnfn isgreaterequal(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isgreaterequal(half x, half y);
 short2 __ovld __cnfn isgreaterequal(half2 x, half2 y);
@@ -10714,14 +10727,14 @@ int3 __ovld __cnfn isless(float3 x, float3 y);
 int4 __ovld __cnfn isless(float4 x, float4 y);
 int8 __ovld __cnfn isless(float8 x, float8 y);
 int16 __ovld __cnfn isless(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isless(double x, double y);
 long2 __ovld __cnfn isless(double2 x, double2 y);
 long3 __ovld __cnfn isless(double3 x, double3 y);
 long4 __ovld __cnfn isless(double4 x, double4 y);
 long8 __ovld __cnfn isless(double8 x, double8 y);
 long16 __ovld __cnfn isless(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isless(half x, half y);
 short2 __ovld __cnfn isless(half2 x, half2 y);
@@ -10740,14 +10753,14 @@ int3 __ovld __cnfn islessequal(float3 x, float3 y);
 int4 __ovld __cnfn islessequal(float4 x, float4 y);
 int8 __ovld __cnfn islessequal(float8 x, float8 y);
 int16 __ovld __cnfn islessequal(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn islessequal(double x, double y);
 long2 __ovld __cnfn islessequal(double2 x, double2 y);
 long3 __ovld __cnfn islessequal(double3 x, double3 y);
 long4 __ovld __cnfn islessequal(double4 x, double4 y);
 long8 __ovld __cnfn islessequal(double8 x, double8 y);
 long16 __ovld __cnfn islessequal(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn islessequal(half x, half y);
 short2 __ovld __cnfn islessequal(half2 x, half2 y);
@@ -10767,14 +10780,14 @@ int3 __ovld __cnfn islessgreater(float3 x, float3 y);
 int4 __ovld __cnfn islessgreater(float4 x, float4 y);
 int8 __ovld __cnfn islessgreater(float8 x, float8 y);
 int16 __ovld __cnfn islessgreater(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn islessgreater(double x, double y);
 long2 __ovld __cnfn islessgreater(double2 x, double2 y);
 long3 __ovld __cnfn islessgreater(double3 x, double3 y);
 long4 __ovld __cnfn islessgreater(double4 x, double4 y);
 long8 __ovld __cnfn islessgreater(double8 x, double8 y);
 long16 __ovld __cnfn islessgreater(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn islessgreater(half x, half y);
 short2 __ovld __cnfn islessgreater(half2 x, half2 y);
@@ -10793,14 +10806,14 @@ int3 __ovld __cnfn isfinite(float3);
 int4 __ovld __cnfn isfinite(float4);
 int8 __ovld __cnfn isfinite(float8);
 int16 __ovld __cnfn isfinite(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isfinite(double);
 long2 __ovld __cnfn isfinite(double2);
 long3 __ovld __cnfn isfinite(double3);
 long4 __ovld __cnfn isfinite(double4);
 long8 __ovld __cnfn isfinite(double8);
 long16 __ovld __cnfn isfinite(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isfinite(half);
 short2 __ovld __cnfn isfinite(half2);
@@ -10819,14 +10832,14 @@ int3 __ovld __cnfn isinf(float3);
 int4 __ovld __cnfn isinf(float4);
 int8 __ovld __cnfn isinf(float8);
 int16 __ovld __cnfn isinf(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isinf(double);
 long2 __ovld __cnfn isinf(double2);
 long3 __ovld __cnfn isinf(double3);
 long4 __ovld __cnfn isinf(double4);
 long8 __ovld __cnfn isinf(double8);
 long16 __ovld __cnfn isinf(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isinf(half);
 short2 __ovld __cnfn isinf(half2);
@@ -10845,14 +10858,14 @@ int3 __ovld __cnfn isnan(float3);
 int4 __ovld __cnfn isnan(float4);
 int8 __ovld __cnfn isnan(float8);
 int16 __ovld __cnfn isnan(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isnan(double);
 long2 __ovld __cnfn isnan(double2);
 long3 __ovld __cnfn isnan(double3);
 long4 __ovld __cnfn isnan(double4);
 long8 __ovld __cnfn isnan(double8);
 long16 __ovld __cnfn isnan(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isnan(half);
 short2 __ovld __cnfn isnan(half2);
@@ -10871,14 +10884,14 @@ int3 __ovld __cnfn isnormal(float3);
 int4 __ovld __cnfn isnormal(float4);
 int8 __ovld __cnfn isnormal(float8);
 int16 __ovld __cnfn isnormal(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isnormal(double);
 long2 __ovld __cnfn isnormal(double2);
 long3 __ovld __cnfn isnormal(double3);
 long4 __ovld __cnfn isnormal(double4);
 long8 __ovld __cnfn isnormal(double8);
 long16 __ovld __cnfn isnormal(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isnormal(half);
 short2 __ovld __cnfn isnormal(half2);
@@ -10899,14 +10912,14 @@ int3 __ovld __cnfn isordered(float3 x, float3 y);
 int4 __ovld __cnfn isordered(float4 x, float4 y);
 int8 __ovld __cnfn isordered(float8 x, float8 y);
 int16 __ovld __cnfn isordered(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isordered(double x, double y);
 long2 __ovld __cnfn isordered(double2 x, double2 y);
 long3 __ovld __cnfn isordered(double3 x, double3 y);
 long4 __ovld __cnfn isordered(double4 x, double4 y);
 long8 __ovld __cnfn isordered(double8 x, double8 y);
 long16 __ovld __cnfn isordered(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isordered(half x, half y);
 short2 __ovld __cnfn isordered(half2 x, half2 y);
@@ -10927,14 +10940,14 @@ int3 __ovld __cnfn isunordered(float3 x, float3 y);
 int4 __ovld __cnfn isunordered(float4 x, float4 y);
 int8 __ovld __cnfn isunordered(float8 x, float8 y);
 int16 __ovld __cnfn isunordered(float16 x, float16 y);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn isunordered(double x, double y);
 long2 __ovld __cnfn isunordered(double2 x, double2 y);
 long3 __ovld __cnfn isunordered(double3 x, double3 y);
 long4 __ovld __cnfn isunordered(double4 x, double4 y);
 long8 __ovld __cnfn isunordered(double8 x, double8 y);
 long16 __ovld __cnfn isunordered(double16 x, double16 y);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn isunordered(half x, half y);
 short2 __ovld __cnfn isunordered(half2 x, half2 y);
@@ -10957,14 +10970,14 @@ int3 __ovld __cnfn signbit(float3);
 int4 __ovld __cnfn signbit(float4);
 int8 __ovld __cnfn signbit(float8);
 int16 __ovld __cnfn signbit(float16);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 int __ovld __cnfn signbit(double);
 long2 __ovld __cnfn signbit(double2);
 long3 __ovld __cnfn signbit(double3);
 long4 __ovld __cnfn signbit(double4);
 long8 __ovld __cnfn signbit(double8);
 long16 __ovld __cnfn signbit(double16);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 int __ovld __cnfn signbit(half);
 short2 __ovld __cnfn signbit(half2);
@@ -11091,14 +11104,14 @@ float3 __ovld __cnfn bitselect(float3 a, float3 b, float3 c);
 float4 __ovld __cnfn bitselect(float4 a, float4 b, float4 c);
 float8 __ovld __cnfn bitselect(float8 a, float8 b, float8 c);
 float16 __ovld __cnfn bitselect(float16 a, float16 b, float16 c);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn bitselect(double a, double b, double c);
 double2 __ovld __cnfn bitselect(double2 a, double2 b, double2 c);
 double3 __ovld __cnfn bitselect(double3 a, double3 b, double3 c);
 double4 __ovld __cnfn bitselect(double4 a, double4 b, double4 c);
 double8 __ovld __cnfn bitselect(double8 a, double8 b, double8 c);
 double16 __ovld __cnfn bitselect(double16 a, double16 b, double16 c);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn bitselect(half a, half b, half c);
 half2 __ovld __cnfn bitselect(half2 a, half2 b, half2 c);
@@ -11231,7 +11244,7 @@ ulong8 __ovld __cnfn select(ulong8 a, ulong8 b, ulong8 c);
 long16 __ovld __cnfn select(long16 a, long16 b, ulong16 c);
 ulong16 __ovld __cnfn select(ulong16 a, ulong16 b, ulong16 c);
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __cnfn select(double a, double b, long c);
 double2 __ovld __cnfn select(double2 a, double2 b, long2 c);
 double3 __ovld __cnfn select(double3 a, double3 b, long3 c);
@@ -11244,7 +11257,7 @@ double3 __ovld __cnfn select(double3 a, double3 b, ulong3 c);
 double4 __ovld __cnfn select(double4 a, double4 b, ulong4 c);
 double8 __ovld __cnfn select(double8 a, double8 b, ulong8 c);
 double16 __ovld __cnfn select(double16 a, double16 b, ulong16 c);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 half __ovld __cnfn select(half a, half b, short c);
 half2 __ovld __cnfn select(half2 a, half2 b, short2 c);
@@ -11323,13 +11336,13 @@ uint16 __ovld vload16(size_t offset, const __constant uint *p);
 long16 __ovld vload16(size_t offset, const __constant long *p);
 ulong16 __ovld vload16(size_t offset, const __constant ulong *p);
 float16 __ovld vload16(size_t offset, const __constant float *p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double2 __ovld vload2(size_t offset, const __constant double *p);
 double3 __ovld vload3(size_t offset, const __constant double *p);
 double4 __ovld vload4(size_t offset, const __constant double *p);
 double8 __ovld vload8(size_t offset, const __constant double *p);
 double16 __ovld vload16(size_t offset, const __constant double *p);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #ifdef cl_khr_fp16
 half __ovld vload(size_t offset, const __constant half *p);
@@ -11340,7 +11353,7 @@ half8 __ovld vload8(size_t offset, const __constant half *p);
 half16 __ovld vload16(size_t offset, const __constant half *p);
 #endif //cl_khr_fp16
 
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 char2 __ovld vload2(size_t offset, const char *p);
 uchar2 __ovld vload2(size_t offset, const uchar *p);
 short2 __ovld vload2(size_t offset, const short *p);
@@ -11387,13 +11400,13 @@ long16 __ovld vload16(size_t offset, const long *p);
 ulong16 __ovld vload16(size_t offset, const ulong *p);
 float16 __ovld vload16(size_t offset, const float *p);
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double2 __ovld vload2(size_t offset, const double *p);
 double3 __ovld vload3(size_t offset, const double *p);
 double4 __ovld vload4(size_t offset, const double *p);
 double8 __ovld vload8(size_t offset, const double *p);
 double16 __ovld vload16(size_t offset, const double *p);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #ifdef cl_khr_fp16
 half __ovld vload(size_t offset, const half *p);
@@ -11403,7 +11416,7 @@ half4 __ovld vload4(size_t offset, const half *p);
 half8 __ovld vload8(size_t offset, const half *p);
 half16 __ovld vload16(size_t offset, const half *p);
 #endif //cl_khr_fp16
-#else
+#endif //__opencl_c_generic_address_space
 char2 __ovld vload2(size_t offset, const __global char *p);
 uchar2 __ovld vload2(size_t offset, const __global uchar *p);
 short2 __ovld vload2(size_t offset, const __global short *p);
@@ -11540,7 +11553,7 @@ long16 __ovld vload16(size_t offset, const __private long *p);
 ulong16 __ovld vload16(size_t offset, const __private ulong *p);
 float16 __ovld vload16(size_t offset, const __private float *p);
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double2 __ovld vload2(size_t offset, const __global double *p);
 double3 __ovld vload3(size_t offset, const __global double *p);
 double4 __ovld vload4(size_t offset, const __global double *p);
@@ -11556,7 +11569,7 @@ double3 __ovld vload3(size_t offset, const __private double *p);
 double4 __ovld vload4(size_t offset, const __private double *p);
 double8 __ovld vload8(size_t offset, const __private double *p);
 double16 __ovld vload16(size_t offset, const __private double *p);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #ifdef cl_khr_fp16
 half __ovld vload(size_t offset, const __global half *p);
@@ -11578,9 +11591,8 @@ half4 __ovld vload4(size_t offset, const __private half *p);
 half8 __ovld vload8(size_t offset, const __private half *p);
 half16 __ovld vload16(size_t offset, const __private half *p);
 #endif //cl_khr_fp16
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 void __ovld vstore2(char2 data, size_t offset, char *p);
 void __ovld vstore2(uchar2 data, size_t offset, uchar *p);
 void __ovld vstore2(short2 data, size_t offset, short *p);
@@ -11626,13 +11638,13 @@ void __ovld vstore16(uint16 data, size_t offset, uint *p);
 void __ovld vstore16(long16 data, size_t offset, long *p);
 void __ovld vstore16(ulong16 data, size_t offset, ulong *p);
 void __ovld vstore16(float16 data, size_t offset, float *p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld vstore2(double2 data, size_t offset, double *p);
 void __ovld vstore3(double3 data, size_t offset, double *p);
 void __ovld vstore4(double4 data, size_t offset, double *p);
 void __ovld vstore8(double8 data, size_t offset, double *p);
 void __ovld vstore16(double16 data, size_t offset, double *p);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 void __ovld vstore(half data, size_t offset, half *p);
 void __ovld vstore2(half2 data, size_t offset, half *p);
@@ -11641,7 +11653,7 @@ void __ovld vstore4(half4 data, size_t offset, half *p);
 void __ovld vstore8(half8 data, size_t offset, half *p);
 void __ovld vstore16(half16 data, size_t offset, half *p);
 #endif //cl_khr_fp16
-#else
+#endif //__opencl_c_generic_address_space
 void __ovld vstore2(char2 data, size_t offset, __global char *p);
 void __ovld vstore2(uchar2 data, size_t offset, __global uchar *p);
 void __ovld vstore2(short2 data, size_t offset, __global short *p);
@@ -11777,7 +11789,7 @@ void __ovld vstore16(uint16 data, size_t offset, __private uint *p);
 void __ovld vstore16(long16 data, size_t offset, __private long *p);
 void __ovld vstore16(ulong16 data, size_t offset, __private ulong *p);
 void __ovld vstore16(float16 data, size_t offset, __private float *p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld vstore2(double2 data, size_t offset, __global double *p);
 void __ovld vstore3(double3 data, size_t offset, __global double *p);
 void __ovld vstore4(double4 data, size_t offset, __global double *p);
@@ -11793,7 +11805,7 @@ void __ovld vstore3(double3 data, size_t offset, __private double *p);
 void __ovld vstore4(double4 data, size_t offset, __private double *p);
 void __ovld vstore8(double8 data, size_t offset, __private double *p);
 void __ovld vstore16(double16 data, size_t offset, __private double *p);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 void __ovld vstore(half data, size_t offset, __global half *p);
 void __ovld vstore2(half2 data, size_t offset, __global half *p);
@@ -11814,7 +11826,6 @@ void __ovld vstore4(half4 data, size_t offset, __private half *p);
 void __ovld vstore8(half8 data, size_t offset, __private half *p);
 void __ovld vstore16(half16 data, size_t offset, __private half *p);
 #endif //cl_khr_fp16
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
  * Read sizeof (half) bytes of data from address
@@ -11825,13 +11836,12 @@ void __ovld vstore16(half16 data, size_t offset, __private half *p);
  * must be 16-bit aligned.
  */
 float __ovld vload_half(size_t offset, const __constant half *p);
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float __ovld vload_half(size_t offset, const half *p);
-#else
+#endif //__opencl_c_generic_address_space
 float __ovld vload_half(size_t offset, const __global half *p);
 float __ovld vload_half(size_t offset, const __local half *p);
 float __ovld vload_half(size_t offset, const __private half *p);
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
  * Read sizeof (halfn) bytes of data from address
@@ -11846,13 +11856,13 @@ float3 __ovld vload_half3(size_t offset, const __constant half *p);
 float4 __ovld vload_half4(size_t offset, const __constant half *p);
 float8 __ovld vload_half8(size_t offset, const __constant half *p);
 float16 __ovld vload_half16(size_t offset, const __constant half *p);
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float2 __ovld vload_half2(size_t offset, const half *p);
 float3 __ovld vload_half3(size_t offset, const half *p);
 float4 __ovld vload_half4(size_t offset, const half *p);
 float8 __ovld vload_half8(size_t offset, const half *p);
 float16 __ovld vload_half16(size_t offset, const half *p);
-#else
+#endif //__opencl_c_generic_address_space
 float2 __ovld vload_half2(size_t offset, const __global half *p);
 float3 __ovld vload_half3(size_t offset, const __global half *p);
 float4 __ovld vload_half4(size_t offset, const __global half *p);
@@ -11868,7 +11878,6 @@ float3 __ovld vload_half3(size_t offset, const __private half *p);
 float4 __ovld vload_half4(size_t offset, const __private half *p);
 float8 __ovld vload_half8(size_t offset, const __private half *p);
 float16 __ovld vload_half16(size_t offset, const __private half *p);
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
  * The float value given by data is first
@@ -11881,20 +11890,20 @@ float16 __ovld vload_half16(size_t offset, const __private half *p);
  * The default current rounding mode is round to
  * nearest even.
  */
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 void __ovld vstore_half(float data, size_t offset, half *p);
 void __ovld vstore_half_rte(float data, size_t offset, half *p);
 void __ovld vstore_half_rtz(float data, size_t offset, half *p);
 void __ovld vstore_half_rtp(float data, size_t offset, half *p);
 void __ovld vstore_half_rtn(float data, size_t offset, half *p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld vstore_half(double data, size_t offset, half *p);
 void __ovld vstore_half_rte(double data, size_t offset, half *p);
 void __ovld vstore_half_rtz(double data, size_t offset, half *p);
 void __ovld vstore_half_rtp(double data, size_t offset, half *p);
 void __ovld vstore_half_rtn(double data, size_t offset, half *p);
-#endif //cl_khr_fp64
-#else
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#endif //__opencl_c_generic_address_space
 void __ovld vstore_half(float data, size_t offset, __global half *p);
 void __ovld vstore_half_rte(float data, size_t offset, __global half *p);
 void __ovld vstore_half_rtz(float data, size_t offset, __global half *p);
@@ -11910,7 +11919,7 @@ void __ovld vstore_half_rte(float data, size_t offset, __private half *p);
 void __ovld vstore_half_rtz(float data, size_t offset, __private half *p);
 void __ovld vstore_half_rtp(float data, size_t offset, __private half *p);
 void __ovld vstore_half_rtn(float data, size_t offset, __private half *p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld vstore_half(double data, size_t offset, __global half *p);
 void __ovld vstore_half_rte(double data, size_t offset, __global half *p);
 void __ovld vstore_half_rtz(double data, size_t offset, __global half *p);
@@ -11926,8 +11935,7 @@ void __ovld vstore_half_rte(double data, size_t offset, __private half *p);
 void __ovld vstore_half_rtz(double data, size_t offset, __private half *p);
 void __ovld vstore_half_rtp(double data, size_t offset, __private half *p);
 void __ovld vstore_half_rtn(double data, size_t offset, __private half *p);
-#endif //cl_khr_fp64
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 /**
  * The floatn value given by data is converted to
@@ -11940,7 +11948,7 @@ void __ovld vstore_half_rtn(double data, size_t offset, __private half *p);
  * The default current rounding mode is round to
  * nearest even.
  */
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 void __ovld vstore_half2(float2 data, size_t offset, half *p);
 void __ovld vstore_half3(float3 data, size_t offset, half *p);
 void __ovld vstore_half4(float4 data, size_t offset, half *p);
@@ -11966,7 +11974,7 @@ void __ovld vstore_half3_rtn(float3 data, size_t offset, half *p);
 void __ovld vstore_half4_rtn(float4 data, size_t offset, half *p);
 void __ovld vstore_half8_rtn(float8 data, size_t offset, half *p);
 void __ovld vstore_half16_rtn(float16 data, size_t offset, half *p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld vstore_half2(double2 data, size_t offset, half *p);
 void __ovld vstore_half3(double3 data, size_t offset, half *p);
 void __ovld vstore_half4(double4 data, size_t offset, half *p);
@@ -11992,8 +12000,8 @@ void __ovld vstore_half3_rtn(double3 data, size_t offset, half *p);
 void __ovld vstore_half4_rtn(double4 data, size_t offset, half *p);
 void __ovld vstore_half8_rtn(double8 data, size_t offset, half *p);
 void __ovld vstore_half16_rtn(double16 data, size_t offset, half *p);
-#endif //cl_khr_fp64
-#else
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#endif //__opencl_c_generic_address_space
 void __ovld vstore_half2(float2 data, size_t offset, __global half *p);
 void __ovld vstore_half3(float3 data, size_t offset, __global half *p);
 void __ovld vstore_half4(float4 data, size_t offset, __global half *p);
@@ -12069,7 +12077,7 @@ void __ovld vstore_half3_rtn(float3 data, size_t offset, __private half *p);
 void __ovld vstore_half4_rtn(float4 data, size_t offset, __private half *p);
 void __ovld vstore_half8_rtn(float8 data, size_t offset, __private half *p);
 void __ovld vstore_half16_rtn(float16 data, size_t offset, __private half *p);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld vstore_half2(double2 data, size_t offset, __global half *p);
 void __ovld vstore_half3(double3 data, size_t offset, __global half *p);
 void __ovld vstore_half4(double4 data, size_t offset, __global half *p);
@@ -12145,8 +12153,7 @@ void __ovld vstore_half3_rtn(double3 data, size_t offset, __private half *p);
 void __ovld vstore_half4_rtn(double4 data, size_t offset, __private half *p);
 void __ovld vstore_half8_rtn(double8 data, size_t offset, __private half *p);
 void __ovld vstore_half16_rtn(double16 data, size_t offset, __private half *p);
-#endif //cl_khr_fp64
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 /**
  * For n = 1, 2, 4, 8 and 16 read sizeof (halfn)
@@ -12167,14 +12174,14 @@ float3 __ovld vloada_half3(size_t offset, const __constant half *p);
 float4 __ovld vloada_half4(size_t offset, const __constant half *p);
 float8 __ovld vloada_half8(size_t offset, const __constant half *p);
 float16 __ovld vloada_half16(size_t offset, const __constant half *p);
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 float __ovld vloada_half(size_t offset, const half *p);
 float2 __ovld vloada_half2(size_t offset, const half *p);
 float3 __ovld vloada_half3(size_t offset, const half *p);
 float4 __ovld vloada_half4(size_t offset, const half *p);
 float8 __ovld vloada_half8(size_t offset, const half *p);
 float16 __ovld vloada_half16(size_t offset, const half *p);
-#else
+#endif //__opencl_c_generic_address_space
 float __ovld vloada_half(size_t offset, const __global half *p);
 float2 __ovld vloada_half2(size_t offset, const __global half *p);
 float3 __ovld vloada_half3(size_t offset, const __global half *p);
@@ -12193,7 +12200,6 @@ float3 __ovld vloada_half3(size_t offset, const __private half *p);
 float4 __ovld vloada_half4(size_t offset, const __private half *p);
 float8 __ovld vloada_half8(size_t offset, const __private half *p);
 float16 __ovld vloada_half16(size_t offset, const __private half *p);
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
  * The floatn value given by data is converted to
@@ -12211,7 +12217,7 @@ float16 __ovld vloada_half16(size_t offset, const __private half *p);
  * mode. The default current rounding mode is
  * round to nearest even.
  */
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 void __ovld vstorea_half(float data, size_t offset, half *p);
 void __ovld vstorea_half2(float2 data, size_t offset, half *p);
 void __ovld vstorea_half3(float3 data, size_t offset, half *p);
@@ -12247,7 +12253,7 @@ void __ovld vstorea_half4_rtn(float4 data, size_t offset, half *p);
 void __ovld vstorea_half8_rtn(float8 data, size_t offset, half *p);
 void __ovld vstorea_half16_rtn(float16 data, size_t offset, half *p);
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld vstorea_half(double data, size_t offset, half *p);
 void __ovld vstorea_half2(double2 data, size_t offset, half *p);
 void __ovld vstorea_half3(double3 data, size_t offset, half *p);
@@ -12282,9 +12288,9 @@ void __ovld vstorea_half3_rtn(double3 data, size_t offset, half *p);
 void __ovld vstorea_half4_rtn(double4 data, size_t offset, half *p);
 void __ovld vstorea_half8_rtn(double8 data, size_t offset, half *p);
 void __ovld vstorea_half16_rtn(double16 data, size_t offset, half *p);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#endif //__opencl_c_generic_address_space
 
-#else
 void __ovld vstorea_half(float data, size_t offset, __global half *p);
 void __ovld vstorea_half2(float2 data, size_t offset, __global half *p);
 void __ovld vstorea_half3(float3 data, size_t offset, __global half *p);
@@ -12390,7 +12396,7 @@ void __ovld vstorea_half4_rtn(float4 data, size_t offset, __private half *p);
 void __ovld vstorea_half8_rtn(float8 data, size_t offset, __private half *p);
 void __ovld vstorea_half16_rtn(float16 data, size_t offset, __private half *p);
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld vstorea_half(double data, size_t offset, __global half *p);
 void __ovld vstorea_half2(double2 data, size_t offset, __global half *p);
 void __ovld vstorea_half3(double3 data, size_t offset, __global half *p);
@@ -12495,8 +12501,7 @@ void __ovld vstorea_half3_rtn(double3 data,size_t offset, __private half *p);
 void __ovld vstorea_half4_rtn(double4 data,size_t offset, __private half *p);
 void __ovld vstorea_half8_rtn(double8 data,size_t offset, __private half *p);
 void __ovld vstorea_half16_rtn(double16 data,size_t offset, __private half *p);
-#endif //cl_khr_fp64
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 // OpenCL v1.1 s6.11.8, v1.2 s6.12.8, v2.0 s6.13.8 - Synchronization Functions
 
@@ -12580,7 +12585,7 @@ void __ovld write_mem_fence(cl_mem_fence_flags flags);
 
 // OpenCL v2.0 s6.13.9 - Address Space Qualifier Functions
 
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_generic_address_space
 cl_mem_fence_flags __ovld get_fence(const void *ptr);
 cl_mem_fence_flags __ovld get_fence(void *ptr);
 
@@ -12591,7 +12596,7 @@ cl_mem_fence_flags __ovld get_fence(void *ptr);
  * where gentype is builtin type or user defined type.
  */
 
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#endif //__opencl_c_generic_address_space
 
 // OpenCL v1.1 s6.11.10, v1.2 s6.12.10, v2.0 s6.13.10 - Async Copies from Global to Local Memory, Local to Global Memory, and Prefetch
 
@@ -12730,7 +12735,7 @@ event_t __ovld async_work_group_copy(__global uint16 *dst, const __local uint16
 event_t __ovld async_work_group_copy(__global long16 *dst, const __local long16 *src, size_t num_elements, event_t event);
 event_t __ovld async_work_group_copy(__global ulong16 *dst, const __local ulong16 *src, size_t num_elements, event_t event);
 event_t __ovld async_work_group_copy(__global float16 *dst, const __local float16 *src, size_t num_elements, event_t event);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 event_t __ovld async_work_group_copy(__local double *dst, const __global double *src, size_t num_elements, event_t event);
 event_t __ovld async_work_group_copy(__local double2 *dst, const __global double2 *src, size_t num_elements, event_t event);
 event_t __ovld async_work_group_copy(__local double3 *dst, const __global double3 *src, size_t num_elements, event_t event);
@@ -12743,7 +12748,7 @@ event_t __ovld async_work_group_copy(__global double3 *dst, const __local double
 event_t __ovld async_work_group_copy(__global double4 *dst, const __local double4 *src, size_t num_elements, event_t event);
 event_t __ovld async_work_group_copy(__global double8 *dst, const __local double8 *src, size_t num_elements, event_t event);
 event_t __ovld async_work_group_copy(__global double16 *dst, const __local double16 *src, size_t num_elements, event_t event);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 event_t __ovld async_work_group_copy(__local half *dst, const __global half *src, size_t num_elements, event_t event);
 event_t __ovld async_work_group_copy(__local half2 *dst, const __global half2 *src, size_t num_elements, event_t event);
@@ -12893,7 +12898,7 @@ event_t __ovld async_work_group_strided_copy(__global uint16 *dst, const __local
 event_t __ovld async_work_group_strided_copy(__global long16 *dst, const __local long16 *src, size_t num_elements, size_t dst_stride, event_t event);
 event_t __ovld async_work_group_strided_copy(__global ulong16 *dst, const __local ulong16 *src, size_t num_elements, size_t dst_stride, event_t event);
 event_t __ovld async_work_group_strided_copy(__global float16 *dst, const __local float16 *src, size_t num_elements, size_t dst_stride, event_t event);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 event_t __ovld async_work_group_strided_copy(__local double *dst, const __global double *src, size_t num_elements, size_t src_stride, event_t event);
 event_t __ovld async_work_group_strided_copy(__local double2 *dst, const __global double2 *src, size_t num_elements, size_t src_stride, event_t event);
 event_t __ovld async_work_group_strided_copy(__local double3 *dst, const __global double3 *src, size_t num_elements, size_t src_stride, event_t event);
@@ -12906,7 +12911,7 @@ event_t __ovld async_work_group_strided_copy(__global double3 *dst, const __loca
 event_t __ovld async_work_group_strided_copy(__global double4 *dst, const __local double4 *src, size_t num_elements, size_t dst_stride, event_t event);
 event_t __ovld async_work_group_strided_copy(__global double8 *dst, const __local double8 *src, size_t num_elements, size_t dst_stride, event_t event);
 event_t __ovld async_work_group_strided_copy(__global double16 *dst, const __local double16 *src, size_t num_elements, size_t dst_stride, event_t event);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 event_t __ovld async_work_group_strided_copy(__local half *dst, const __global half *src, size_t num_elements, size_t src_stride, event_t event);
 event_t __ovld async_work_group_strided_copy(__local half2 *dst, const __global half2 *src, size_t num_elements, size_t src_stride, event_t event);
@@ -12996,14 +13001,14 @@ void __ovld prefetch(const __global uint16 *p, size_t num_elements);
 void __ovld prefetch(const __global long16 *p, size_t num_elements);
 void __ovld prefetch(const __global ulong16 *p, size_t num_elements);
 void __ovld prefetch(const __global float16 *p, size_t num_elements);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld prefetch(const __global double *p, size_t num_elements);
 void __ovld prefetch(const __global double2 *p, size_t num_elements);
 void __ovld prefetch(const __global double3 *p, size_t num_elements);
 void __ovld prefetch(const __global double4 *p, size_t num_elements);
 void __ovld prefetch(const __global double8 *p, size_t num_elements);
 void __ovld prefetch(const __global double16 *p, size_t num_elements);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #ifdef cl_khr_fp16
 void __ovld prefetch(const __global half *p, size_t num_elements);
 void __ovld prefetch(const __global half2 *p, size_t num_elements);
@@ -13026,9 +13031,11 @@ void __ovld prefetch(const __global half16 *p, size_t num_elements);
  * pointed by p. The function returns old.
  */
 int __ovld atomic_add(volatile __global int *p, int val);
-unsigned int __ovld atomic_add(volatile __global unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_add(volatile __global unsigned int *p,
+                               unsigned int val);
 int __ovld atomic_add(volatile __local int *p, int val);
-unsigned int __ovld atomic_add(volatile __local unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_add(volatile __local unsigned int *p,
+                               unsigned int val);
 #ifdef __OPENCL_CPP_VERSION__
 int __ovld atomic_add(volatile int *p, int val);
 unsigned int __ovld atomic_add(volatile unsigned int *p, unsigned int val);
@@ -13056,9 +13063,11 @@ unsigned long __ovld atom_add(volatile __local unsigned long *p, unsigned long v
  * returns old.
  */
 int __ovld atomic_sub(volatile __global int *p, int val);
-unsigned int __ovld atomic_sub(volatile __global unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_sub(volatile __global unsigned int *p,
+                               unsigned int val);
 int __ovld atomic_sub(volatile __local int *p, int val);
-unsigned int __ovld atomic_sub(volatile __local unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_sub(volatile __local unsigned int *p,
+                               unsigned int val);
 #ifdef __OPENCL_CPP_VERSION__
 int __ovld atomic_sub(volatile int *p, int val);
 unsigned int __ovld atomic_sub(volatile unsigned int *p, unsigned int val);
@@ -13086,9 +13095,11 @@ unsigned long __ovld atom_sub(volatile __local unsigned long *p, unsigned long v
  * value.
  */
 int __ovld atomic_xchg(volatile __global int *p, int val);
-unsigned int __ovld atomic_xchg(volatile __global unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_xchg(volatile __global unsigned int *p,
+                                unsigned int val);
 int __ovld atomic_xchg(volatile __local int *p, int val);
-unsigned int __ovld atomic_xchg(volatile __local unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_xchg(volatile __local unsigned int *p,
+                                unsigned int val);
 float __ovld atomic_xchg(volatile __global float *p, float val);
 float __ovld atomic_xchg(volatile __local float *p, float val);
 #ifdef __OPENCL_CPP_VERSION__
@@ -13183,12 +13194,15 @@ unsigned long __ovld atom_dec(volatile __local unsigned long *p);
  * returns old.
  */
 int __ovld atomic_cmpxchg(volatile __global int *p, int cmp, int val);
-unsigned int __ovld atomic_cmpxchg(volatile __global unsigned int *p, unsigned int cmp, unsigned int val);
+unsigned int __ovld atomic_cmpxchg(volatile __global unsigned int *p,
+                                   unsigned int cmp, unsigned int val);
 int __ovld atomic_cmpxchg(volatile __local int *p, int cmp, int val);
-unsigned int __ovld atomic_cmpxchg(volatile __local unsigned int *p, unsigned int cmp, unsigned int val);
+unsigned int __ovld atomic_cmpxchg(volatile __local unsigned int *p,
+                                   unsigned int cmp, unsigned int val);
 #ifdef __OPENCL_CPP_VERSION__
 int __ovld atomic_cmpxchg(volatile int *p, int cmp, int val);
-unsigned int __ovld atomic_cmpxchg(volatile unsigned int *p, unsigned int cmp, unsigned int val);
+unsigned int __ovld atomic_cmpxchg(volatile unsigned int *p, unsigned int cmp,
+                                   unsigned int val);
 #endif
 
 #if defined(cl_khr_global_int32_base_atomics)
@@ -13215,9 +13229,11 @@ unsigned long __ovld atom_cmpxchg(volatile __local unsigned long *p, unsigned lo
  * returns old.
  */
 int __ovld atomic_min(volatile __global int *p, int val);
-unsigned int __ovld atomic_min(volatile __global unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_min(volatile __global unsigned int *p,
+                               unsigned int val);
 int __ovld atomic_min(volatile __local int *p, int val);
-unsigned int __ovld atomic_min(volatile __local unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_min(volatile __local unsigned int *p,
+                               unsigned int val);
 #ifdef __OPENCL_CPP_VERSION__
 int __ovld atomic_min(volatile int *p, int val);
 unsigned int __ovld atomic_min(volatile unsigned int *p, unsigned int val);
@@ -13247,9 +13263,11 @@ unsigned long __ovld atom_min(volatile __local unsigned long *p, unsigned long v
  * returns old.
  */
 int __ovld atomic_max(volatile __global int *p, int val);
-unsigned int __ovld atomic_max(volatile __global unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_max(volatile __global unsigned int *p,
+                               unsigned int val);
 int __ovld atomic_max(volatile __local int *p, int val);
-unsigned int __ovld atomic_max(volatile __local unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_max(volatile __local unsigned int *p,
+                               unsigned int val);
 #ifdef __OPENCL_CPP_VERSION__
 int __ovld atomic_max(volatile int *p, int val);
 unsigned int __ovld atomic_max(volatile unsigned int *p, unsigned int val);
@@ -13278,9 +13296,11 @@ unsigned long __ovld atom_max(volatile __local unsigned long *p, unsigned long v
  * pointed by p. The function returns old.
  */
 int __ovld atomic_and(volatile __global int *p, int val);
-unsigned int __ovld atomic_and(volatile __global unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_and(volatile __global unsigned int *p,
+                               unsigned int val);
 int __ovld atomic_and(volatile __local int *p, int val);
-unsigned int __ovld atomic_and(volatile __local unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_and(volatile __local unsigned int *p,
+                               unsigned int val);
 #ifdef __OPENCL_CPP_VERSION__
 int __ovld atomic_and(volatile int *p, int val);
 unsigned int __ovld atomic_and(volatile unsigned int *p, unsigned int val);
@@ -13309,9 +13329,11 @@ unsigned long __ovld atom_and(volatile __local unsigned long *p, unsigned long v
  * pointed by p. The function returns old.
  */
 int __ovld atomic_or(volatile __global int *p, int val);
-unsigned int __ovld atomic_or(volatile __global unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_or(volatile __global unsigned int *p,
+                              unsigned int val);
 int __ovld atomic_or(volatile __local int *p, int val);
-unsigned int __ovld atomic_or(volatile __local unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_or(volatile __local unsigned int *p,
+                              unsigned int val);
 #ifdef __OPENCL_CPP_VERSION__
 int __ovld atomic_or(volatile int *p, int val);
 unsigned int __ovld atomic_or(volatile unsigned int *p, unsigned int val);
@@ -13340,9 +13362,11 @@ unsigned long __ovld atom_or(volatile __local unsigned long *p, unsigned long va
  * pointed by p. The function returns old.
  */
 int __ovld atomic_xor(volatile __global int *p, int val);
-unsigned int __ovld atomic_xor(volatile __global unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_xor(volatile __global unsigned int *p,
+                               unsigned int val);
 int __ovld atomic_xor(volatile __local int *p, int val);
-unsigned int __ovld atomic_xor(volatile __local unsigned int *p, unsigned int val);
+unsigned int __ovld atomic_xor(volatile __local unsigned int *p,
+                               unsigned int val);
 #ifdef __OPENCL_CPP_VERSION__
 int __ovld atomic_xor(volatile int *p, int val);
 unsigned int __ovld atomic_xor(volatile unsigned int *p, unsigned int val);
@@ -13380,120 +13404,78 @@ unsigned long __ovld atom_xor(volatile __local unsigned long *p, unsigned long v
 #endif
 
 // atomic_init()
+#ifdef __opencl_c_generic_address_space
 void __ovld atomic_init(volatile atomic_int *object, int value);
 void __ovld atomic_init(volatile atomic_uint *object, uint value);
 void __ovld atomic_init(volatile atomic_float *object, float value);
 #if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
 void __ovld atomic_init(volatile atomic_long *object, long value);
 void __ovld atomic_init(volatile atomic_ulong *object, ulong value);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld atomic_init(volatile atomic_double *object, double value);
-#endif //cl_khr_fp64
-#endif
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+void __ovld atomic_init(volatile atomic_int __global *object, int value);
+void __ovld atomic_init(volatile atomic_int __local *object, int value);
+void __ovld atomic_init(volatile atomic_uint __global *object, uint value);
+void __ovld atomic_init(volatile atomic_uint __local *object, uint value);
+void __ovld atomic_init(volatile atomic_float __global *object, float value);
+void __ovld atomic_init(volatile atomic_float __local *object, float value);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+void __ovld atomic_init(volatile atomic_long __global *object, long value);
+void __ovld atomic_init(volatile atomic_long __local *object, long value);
+void __ovld atomic_init(volatile atomic_ulong __global *object, ulong value);
+void __ovld atomic_init(volatile atomic_ulong __local *object, ulong value);
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+void __ovld atomic_init(volatile atomic_double __global *object, double value);
+void __ovld atomic_init(volatile atomic_double __local *object, double value);
+#endif // cl_khr_fp64
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
 
 // atomic_work_item_fence()
-void __ovld atomic_work_item_fence(cl_mem_fence_flags flags, memory_order order, memory_scope scope);
+void __ovld atomic_work_item_fence(cl_mem_fence_flags flags, memory_order order,
+                                   memory_scope scope);
 
 // atomic_fetch()
-
+#if defined(__opencl_c_atomic_scope_device) &&                              \
+    defined(__opencl_c_atomic_order_seq_cst)
+#ifdef __opencl_c_generic_address_space
 int __ovld atomic_fetch_add(volatile atomic_int *object, int operand);
-int __ovld atomic_fetch_add_explicit(volatile atomic_int *object, int operand, memory_order order);
-int __ovld atomic_fetch_add_explicit(volatile atomic_int *object, int operand, memory_order order, memory_scope scope);
 uint __ovld atomic_fetch_add(volatile atomic_uint *object, uint operand);
-uint __ovld atomic_fetch_add_explicit(volatile atomic_uint *object, uint operand, memory_order order);
-uint __ovld atomic_fetch_add_explicit(volatile atomic_uint *object, uint operand, memory_order order, memory_scope scope);
 int __ovld atomic_fetch_sub(volatile atomic_int *object, int operand);
-int __ovld atomic_fetch_sub_explicit(volatile atomic_int *object, int operand, memory_order order);
-int __ovld atomic_fetch_sub_explicit(volatile atomic_int *object, int operand, memory_order order, memory_scope scope);
 uint __ovld atomic_fetch_sub(volatile atomic_uint *object, uint operand);
-uint __ovld atomic_fetch_sub_explicit(volatile atomic_uint *object, uint operand, memory_order order);
-uint __ovld atomic_fetch_sub_explicit(volatile atomic_uint *object, uint operand, memory_order order, memory_scope scope);
 int __ovld atomic_fetch_or(volatile atomic_int *object, int operand);
-int __ovld atomic_fetch_or_explicit(volatile atomic_int *object, int operand, memory_order order);
-int __ovld atomic_fetch_or_explicit(volatile atomic_int *object, int operand, memory_order order, memory_scope scope);
 uint __ovld atomic_fetch_or(volatile atomic_uint *object, uint operand);
-uint __ovld atomic_fetch_or_explicit(volatile atomic_uint *object, uint operand, memory_order order);
-uint __ovld atomic_fetch_or_explicit(volatile atomic_uint *object, uint operand, memory_order order, memory_scope scope);
 int __ovld atomic_fetch_xor(volatile atomic_int *object, int operand);
-int __ovld atomic_fetch_xor_explicit(volatile atomic_int *object, int operand, memory_order order);
-int __ovld atomic_fetch_xor_explicit(volatile atomic_int *object, int operand, memory_order order, memory_scope scope);
 uint __ovld atomic_fetch_xor(volatile atomic_uint *object, uint operand);
-uint __ovld atomic_fetch_xor_explicit(volatile atomic_uint *object, uint operand, memory_order order);
-uint __ovld atomic_fetch_xor_explicit(volatile atomic_uint *object, uint operand, memory_order order, memory_scope scope);
 int __ovld atomic_fetch_and(volatile atomic_int *object, int operand);
-int __ovld atomic_fetch_and_explicit(volatile atomic_int *object, int operand, memory_order order);
-int __ovld atomic_fetch_and_explicit(volatile atomic_int *object, int operand, memory_order order, memory_scope scope);
 uint __ovld atomic_fetch_and(volatile atomic_uint *object, uint operand);
-uint __ovld atomic_fetch_and_explicit(volatile atomic_uint *object, uint operand, memory_order order);
-uint __ovld atomic_fetch_and_explicit(volatile atomic_uint *object, uint operand, memory_order order, memory_scope scope);
 int __ovld atomic_fetch_min(volatile atomic_int *object, int operand);
-int __ovld atomic_fetch_min_explicit(volatile atomic_int *object, int operand, memory_order order);
-int __ovld atomic_fetch_min_explicit(volatile atomic_int *object, int operand, memory_order order, memory_scope scope);
 uint __ovld atomic_fetch_min(volatile atomic_uint *object, uint operand);
-uint __ovld atomic_fetch_min_explicit(volatile atomic_uint *object, uint operand, memory_order order);
-uint __ovld atomic_fetch_min_explicit(volatile atomic_uint *object, uint operand, memory_order order, memory_scope scope);
-uint __ovld atomic_fetch_min(volatile atomic_uint *object, int operand);
-uint __ovld atomic_fetch_min_explicit(volatile atomic_uint *object, int operand, memory_order order);
-uint __ovld atomic_fetch_min_explicit(volatile atomic_uint *object, int operand, memory_order order, memory_scope scope);
 int __ovld atomic_fetch_max(volatile atomic_int *object, int operand);
-int __ovld atomic_fetch_max_explicit(volatile atomic_int *object, int operand, memory_order order);
-int __ovld atomic_fetch_max_explicit(volatile atomic_int *object, int operand, memory_order order, memory_scope scope);
 uint __ovld atomic_fetch_max(volatile atomic_uint *object, uint operand);
-uint __ovld atomic_fetch_max_explicit(volatile atomic_uint *object, uint operand, memory_order order);
-uint __ovld atomic_fetch_max_explicit(volatile atomic_uint *object, uint operand, memory_order order, memory_scope scope);
-uint __ovld atomic_fetch_max(volatile atomic_uint *object, int operand);
-uint __ovld atomic_fetch_max_explicit(volatile atomic_uint *object, int operand, memory_order order);
-uint __ovld atomic_fetch_max_explicit(volatile atomic_uint *object, int operand, memory_order order, memory_scope scope);
 
 #if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
 long __ovld atomic_fetch_add(volatile atomic_long *object, long operand);
-long __ovld atomic_fetch_add_explicit(volatile atomic_long *object, long operand, memory_order order);
-long __ovld atomic_fetch_add_explicit(volatile atomic_long *object, long operand, memory_order order, memory_scope scope);
 ulong __ovld atomic_fetch_add(volatile atomic_ulong *object, ulong operand);
-ulong __ovld atomic_fetch_add_explicit(volatile atomic_ulong *object, ulong operand, memory_order order);
-ulong __ovld atomic_fetch_add_explicit(volatile atomic_ulong *object, ulong operand, memory_order order, memory_scope scope);
 long __ovld atomic_fetch_sub(volatile atomic_long *object, long operand);
-long __ovld atomic_fetch_sub_explicit(volatile atomic_long *object, long operand, memory_order order);
-long __ovld atomic_fetch_sub_explicit(volatile atomic_long *object, long operand, memory_order order, memory_scope scope);
 ulong __ovld atomic_fetch_sub(volatile atomic_ulong *object, ulong operand);
-ulong __ovld atomic_fetch_sub_explicit(volatile atomic_ulong *object, ulong operand, memory_order order);
-ulong __ovld atomic_fetch_sub_explicit(volatile atomic_ulong *object, ulong operand, memory_order order, memory_scope scope);
 long __ovld atomic_fetch_or(volatile atomic_long *object, long operand);
-long __ovld atomic_fetch_or_explicit(volatile atomic_long *object, long operand, memory_order order);
-long __ovld atomic_fetch_or_explicit(volatile atomic_long *object, long operand, memory_order order, memory_scope scope);
 ulong __ovld atomic_fetch_or(volatile atomic_ulong *object, ulong operand);
-ulong __ovld atomic_fetch_or_explicit(volatile atomic_ulong *object, ulong operand, memory_order order);
-ulong __ovld atomic_fetch_or_explicit(volatile atomic_ulong *object, ulong operand, memory_order order, memory_scope scope);
 long __ovld atomic_fetch_xor(volatile atomic_long *object, long operand);
-long __ovld atomic_fetch_xor_explicit(volatile atomic_long *object, long operand, memory_order order);
-long __ovld atomic_fetch_xor_explicit(volatile atomic_long *object, long operand, memory_order order, memory_scope scope);
 ulong __ovld atomic_fetch_xor(volatile atomic_ulong *object, ulong operand);
-ulong __ovld atomic_fetch_xor_explicit(volatile atomic_ulong *object, ulong operand, memory_order order);
-ulong __ovld atomic_fetch_xor_explicit(volatile atomic_ulong *object, ulong operand, memory_order order, memory_scope scope);
 long __ovld atomic_fetch_and(volatile atomic_long *object, long operand);
-long __ovld atomic_fetch_and_explicit(volatile atomic_long *object, long operand, memory_order order);
-long __ovld atomic_fetch_and_explicit(volatile atomic_long *object, long operand, memory_order order, memory_scope scope);
 ulong __ovld atomic_fetch_and(volatile atomic_ulong *object, ulong operand);
-ulong __ovld atomic_fetch_and_explicit(volatile atomic_ulong *object, ulong operand, memory_order order);
-ulong __ovld atomic_fetch_and_explicit(volatile atomic_ulong *object, ulong operand, memory_order order, memory_scope scope);
 long __ovld atomic_fetch_min(volatile atomic_long *object, long operand);
-long __ovld atomic_fetch_min_explicit(volatile atomic_long *object, long operand, memory_order order);
-long __ovld atomic_fetch_min_explicit(volatile atomic_long *object, long operand, memory_order order, memory_scope scope);
 ulong __ovld atomic_fetch_min(volatile atomic_ulong *object, ulong operand);
-ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong *object, ulong operand, memory_order order);
-ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong *object, ulong operand, memory_order order, memory_scope scope);
-ulong __ovld atomic_fetch_min(volatile atomic_ulong *object, long operand);
-ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong *object, long operand, memory_order order);
-ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong *object, long operand, memory_order order, memory_scope scope);
 long __ovld atomic_fetch_max(volatile atomic_long *object, long operand);
-long __ovld atomic_fetch_max_explicit(volatile atomic_long *object, long operand, memory_order order);
-long __ovld atomic_fetch_max_explicit(volatile atomic_long *object, long operand, memory_order order, memory_scope scope);
 ulong __ovld atomic_fetch_max(volatile atomic_ulong *object, ulong operand);
-ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong *object, ulong operand, memory_order order);
-ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong *object, ulong operand, memory_order order, memory_scope scope);
-ulong __ovld atomic_fetch_max(volatile atomic_ulong *object, long operand);
-ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong *object, long operand, memory_order order);
-ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong *object, long operand, memory_order order, memory_scope scope);
 #endif //defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
 
 // OpenCL v2.0 s6.13.11.7.5:
@@ -13501,196 +13483,2239 @@ ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong *object, long opera
 // or/xor/and/min/max: atomic type argument can be intptr_t/uintptr_t, value type argument can be intptr_t/uintptr_t.
 
 #if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
-uintptr_t __ovld atomic_fetch_add(volatile atomic_uintptr_t *object, ptrdiff_t operand);
-uintptr_t __ovld atomic_fetch_add_explicit(volatile atomic_uintptr_t *object, ptrdiff_t operand, memory_order order);
-uintptr_t __ovld atomic_fetch_add_explicit(volatile atomic_uintptr_t *object, ptrdiff_t operand, memory_order order, memory_scope scope);
-uintptr_t __ovld atomic_fetch_sub(volatile atomic_uintptr_t *object, ptrdiff_t operand);
-uintptr_t __ovld atomic_fetch_sub_explicit(volatile atomic_uintptr_t *object, ptrdiff_t operand, memory_order order);
-uintptr_t __ovld atomic_fetch_sub_explicit(volatile atomic_uintptr_t *object, ptrdiff_t operand, memory_order order, memory_scope scope);
-
-uintptr_t __ovld atomic_fetch_or(volatile atomic_uintptr_t *object, intptr_t operand);
-uintptr_t __ovld atomic_fetch_or_explicit(volatile atomic_uintptr_t *object, intptr_t operand, memory_order order);
-uintptr_t __ovld atomic_fetch_or_explicit(volatile atomic_uintptr_t *object, intptr_t operand, memory_order order, memory_scope scope);
-uintptr_t __ovld atomic_fetch_xor(volatile atomic_uintptr_t *object, intptr_t operand);
-uintptr_t __ovld atomic_fetch_xor_explicit(volatile atomic_uintptr_t *object, intptr_t operand, memory_order order);
-uintptr_t __ovld atomic_fetch_xor_explicit(volatile atomic_uintptr_t *object, intptr_t operand, memory_order order, memory_scope scope);
-uintptr_t __ovld atomic_fetch_and(volatile atomic_uintptr_t *object, intptr_t operand);
-uintptr_t __ovld atomic_fetch_and_explicit(volatile atomic_uintptr_t *object, intptr_t operand, memory_order order);
-uintptr_t __ovld atomic_fetch_and_explicit(volatile atomic_uintptr_t *object, intptr_t operand, memory_order order, memory_scope scope);
-uintptr_t __ovld atomic_fetch_min(volatile atomic_uintptr_t *object, intptr_t opermax);
-uintptr_t __ovld atomic_fetch_min_explicit(volatile atomic_uintptr_t *object, intptr_t opermax, memory_order minder);
-uintptr_t __ovld atomic_fetch_min_explicit(volatile atomic_uintptr_t *object, intptr_t opermax, memory_order minder, memory_scope scope);
-uintptr_t __ovld atomic_fetch_max(volatile atomic_uintptr_t *object, intptr_t opermax);
-uintptr_t __ovld atomic_fetch_max_explicit(volatile atomic_uintptr_t *object, intptr_t opermax, memory_order minder);
-uintptr_t __ovld atomic_fetch_max_explicit(volatile atomic_uintptr_t *object, intptr_t opermax, memory_order minder, memory_scope scope);
-
-intptr_t __ovld atomic_fetch_or(volatile atomic_intptr_t *object, uintptr_t operand);
-intptr_t __ovld atomic_fetch_or_explicit(volatile atomic_intptr_t *object, uintptr_t operand, memory_order order);
-intptr_t __ovld atomic_fetch_or_explicit(volatile atomic_intptr_t *object, uintptr_t operand, memory_order order, memory_scope scope);
-intptr_t __ovld atomic_fetch_xor(volatile atomic_intptr_t *object, uintptr_t operand);
-intptr_t __ovld atomic_fetch_xor_explicit(volatile atomic_intptr_t *object, uintptr_t operand, memory_order order);
-intptr_t __ovld atomic_fetch_xor_explicit(volatile atomic_intptr_t *object, uintptr_t operand, memory_order order, memory_scope scope);
-intptr_t __ovld atomic_fetch_and(volatile atomic_intptr_t *object, uintptr_t operand);
-intptr_t __ovld atomic_fetch_and_explicit(volatile atomic_intptr_t *object, uintptr_t operand, memory_order order);
-intptr_t __ovld atomic_fetch_and_explicit(volatile atomic_intptr_t *object, uintptr_t operand, memory_order order, memory_scope scope);
-intptr_t __ovld atomic_fetch_min(volatile atomic_intptr_t *object, uintptr_t opermax);
-intptr_t __ovld atomic_fetch_min_explicit(volatile atomic_intptr_t *object, uintptr_t opermax, memory_order minder);
-intptr_t __ovld atomic_fetch_min_explicit(volatile atomic_intptr_t *object, uintptr_t opermax, memory_order minder, memory_scope scope);
-intptr_t __ovld atomic_fetch_max(volatile atomic_intptr_t *object, uintptr_t opermax);
-intptr_t __ovld atomic_fetch_max_explicit(volatile atomic_intptr_t *object, uintptr_t opermax, memory_order minder);
-intptr_t __ovld atomic_fetch_max_explicit(volatile atomic_intptr_t *object, uintptr_t opermax, memory_order minder, memory_scope scope);
+uintptr_t __ovld atomic_fetch_add(volatile atomic_uintptr_t *object,
+                                  ptrdiff_t operand);
+uintptr_t __ovld atomic_fetch_sub(volatile atomic_uintptr_t *object,
+                                  ptrdiff_t operand);
+
+uintptr_t __ovld atomic_fetch_or(volatile atomic_uintptr_t *object,
+                                 intptr_t operand);
+uintptr_t __ovld atomic_fetch_xor(volatile atomic_uintptr_t *object,
+                                  intptr_t operand);
+uintptr_t __ovld atomic_fetch_and(volatile atomic_uintptr_t *object,
+                                  intptr_t operand);
+uintptr_t __ovld atomic_fetch_min(volatile atomic_uintptr_t *object,
+                                  intptr_t opermax);
+uintptr_t __ovld atomic_fetch_max(volatile atomic_uintptr_t *object,
+                                  intptr_t opermax);
+
+intptr_t __ovld atomic_fetch_or(volatile atomic_intptr_t *object,
+                                uintptr_t operand);
+intptr_t __ovld atomic_fetch_xor(volatile atomic_intptr_t *object,
+                                 uintptr_t operand);
+intptr_t __ovld atomic_fetch_and(volatile atomic_intptr_t *object,
+                                 uintptr_t operand);
+intptr_t __ovld atomic_fetch_min(volatile atomic_intptr_t *object,
+                                 uintptr_t opermax);
+intptr_t __ovld atomic_fetch_max(volatile atomic_intptr_t *object,
+                                 uintptr_t opermax);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // __opencl_c_generic_address_space
+
+#if(__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+int __ovld atomic_fetch_add(volatile atomic_int __global *object, int operand);
+uint __ovld atomic_fetch_add(volatile atomic_uint __local *object,
+                             uint operand);
+int __ovld atomic_fetch_sub(volatile atomic_int __global *object, int operand);
+int __ovld atomic_fetch_sub(volatile atomic_int __local *object, int operand);
+uint __ovld atomic_fetch_sub(volatile atomic_uint __local *object,
+                             uint operand);
+uint __ovld atomic_fetch_sub(volatile atomic_uint __global *object,
+                             uint operand);
+int __ovld atomic_fetch_or(volatile atomic_int __global *object, int operand);
+uint __ovld atomic_fetch_sub(volatile atomic_uint __local *object,
+                             uint operand);
+uint __ovld atomic_fetch_or(volatile atomic_uint __global *object,
+                            uint operand);
+uint __ovld atomic_fetch_or(volatile atomic_uint __local *object, uint operand);
+int __ovld atomic_fetch_xor(volatile atomic_int __global *object, int operand);
+int __ovld atomic_fetch_xor(volatile atomic_int __local *object, int operand);
+uint __ovld atomic_fetch_xor(volatile atomic_uint __global *object,
+                             uint operand);
+uint __ovld atomic_fetch_xor(volatile atomic_uint __local *object,
+                             uint operand);
+int __ovld atomic_fetch_and(volatile atomic_int __global *object, int operand);
+int __ovld atomic_fetch_and(volatile atomic_int __local *object, int operand);
+uint __ovld atomic_fetch_and(volatile atomic_uint __global *object,
+                             uint operand);
+uint __ovld atomic_fetch_and(volatile atomic_uint __local *object,
+                             uint operand);
+int __ovld atomic_fetch_min(volatile atomic_int __global *object, int operand);
+int __ovld atomic_fetch_min(volatile atomic_int __local *object, int operand);
+uint __ovld atomic_fetch_min(volatile atomic_uint __global *object,
+                             uint operand);
+uint __ovld atomic_fetch_min(volatile atomic_uint __local *object,
+                             uint operand);
+int __ovld atomic_fetch_max(volatile atomic_int __global *object, int operand);
+int __ovld atomic_fetch_max(volatile atomic_int __local *object, int operand);
+uint __ovld atomic_fetch_max(volatile atomic_uint __global *object,
+                             uint operand);
+uint __ovld atomic_fetch_max(volatile atomic_uint __local *object,
+                             uint operand);
+
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+long __ovld atomic_fetch_add(volatile atomic_long __global *object,
+                             long operand);
+long __ovld atomic_fetch_add(volatile atomic_long __local *object,
+                             long operand);
+ulong __ovld atomic_fetch_add(volatile atomic_ulong __global *object,
+                              ulong operand);
+ulong __ovld atomic_fetch_add(volatile atomic_ulong __local *object,
+                              ulong operand);
+long __ovld atomic_fetch_sub(volatile atomic_long __global *object,
+                             long operand);
+long __ovld atomic_fetch_sub(volatile atomic_long __local *object,
+                             long operand);
+ulong __ovld atomic_fetch_sub(volatile atomic_ulong __global *object,
+                              ulong operand);
+ulong __ovld atomic_fetch_sub(volatile atomic_ulong __local *object,
+                              ulong operand);
+long __ovld atomic_fetch_or(volatile atomic_long __global *object,
+                            long operand);
+long __ovld atomic_fetch_or(volatile atomic_long __local *object, long operand);
+ulong __ovld atomic_fetch_or(volatile atomic_ulong __global *object,
+                             ulong operand);
+ulong __ovld atomic_fetch_or(volatile atomic_ulong __local *object,
+                             ulong operand);
+long __ovld atomic_fetch_xor(volatile atomic_long __global *object,
+                             long operand);
+long __ovld atomic_fetch_xor(volatile atomic_long __local *object,
+                             long operand);
+ulong __ovld atomic_fetch_xor(volatile atomic_ulong __global *object,
+                              ulong operand);
+ulong __ovld atomic_fetch_xor(volatile atomic_ulong __local *object,
+                              ulong operand);
+long __ovld atomic_fetch_and(volatile atomic_long __global *object,
+                             long operand);
+long __ovld atomic_fetch_and(volatile atomic_long __local *object,
+                             long operand);
+ulong __ovld atomic_fetch_and(volatile atomic_ulong __global *object,
+                              ulong operand);
+ulong __ovld atomic_fetch_and(volatile atomic_ulong __local *object,
+                              ulong operand);
+long __ovld atomic_fetch_min(volatile atomic_long __global *object,
+                             long operand);
+long __ovld atomic_fetch_min(volatile atomic_long __local *object,
+                             long operand);
+ulong __ovld atomic_fetch_min(volatile atomic_ulong __global *object,
+                              ulong operand);
+ulong __ovld atomic_fetch_min(volatile atomic_ulong __local *object,
+                              ulong operand);
+long __ovld atomic_fetch_max(volatile atomic_long __global *object,
+                             long operand);
+long __ovld atomic_fetch_max(volatile atomic_long __local *object,
+                             long operand);
+ulong __ovld atomic_fetch_max(volatile atomic_ulong __global *object,
+                              ulong operand);
+ulong __ovld atomic_fetch_max(volatile atomic_ulong __local *object,
+                              ulong operand);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+
+// OpenCL v2.0 s6.13.11.7.5:
+// add/sub: atomic type argument can be uintptr_t/intptr_t, value type argument
+// can be ptrdiff_t. or/xor/and/min/max: atomic type argument can be
+// intptr_t/uintptr_t, value type argument can be intptr_t/uintptr_t.
+
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+uintptr_t __ovld atomic_fetch_add(volatile atomic_uintptr_t __global *object,
+                                  ptrdiff_t operand);
+uintptr_t __ovld atomic_fetch_add(volatile atomic_uintptr_t __local *object,
+                                  ptrdiff_t operand);
+uintptr_t __ovld atomic_fetch_sub(volatile atomic_uintptr_t __global *object,
+                                  ptrdiff_t operand);
+uintptr_t __ovld atomic_fetch_sub(volatile atomic_uintptr_t __local *object,
+                                  ptrdiff_t operand);
+
+uintptr_t __ovld atomic_fetch_or(volatile atomic_uintptr_t __global *object,
+                                 intptr_t operand);
+uintptr_t __ovld atomic_fetch_or(volatile atomic_uintptr_t __local *object,
+                                 intptr_t operand);
+uintptr_t __ovld atomic_fetch_xor(volatile atomic_uintptr_t __global *object,
+                                  intptr_t operand);
+uintptr_t __ovld atomic_fetch_xor(volatile atomic_uintptr_t __local *object,
+                                  intptr_t operand);
+uintptr_t __ovld atomic_fetch_and(volatile atomic_uintptr_t __global *object,
+                                  intptr_t operand);
+uintptr_t __ovld atomic_fetch_and(volatile atomic_uintptr_t __local *object,
+                                  intptr_t operand);
+uintptr_t __ovld atomic_fetch_min(volatile atomic_uintptr_t __global *object,
+                                  intptr_t opermax);
+uintptr_t __ovld atomic_fetch_min(volatile atomic_uintptr_t __local *object,
+                                  intptr_t opermax);
+uintptr_t __ovld atomic_fetch_max(volatile atomic_uintptr_t __global *object,
+                                  intptr_t opermax);
+uintptr_t __ovld atomic_fetch_max(volatile atomic_uintptr_t __local *object,
+                                  intptr_t opermax);
+
+intptr_t __ovld atomic_fetch_or(volatile atomic_intptr_t __global *object,
+                                uintptr_t operand);
+intptr_t __ovld atomic_fetch_or(volatile atomic_intptr_t __local *object,
+                                uintptr_t operand);
+intptr_t __ovld atomic_fetch_xor(volatile atomic_intptr_t __global *object,
+                                 uintptr_t operand);
+intptr_t __ovld atomic_fetch_xor(volatile atomic_intptr_t __local *object,
+                                 uintptr_t operand);
+intptr_t __ovld atomic_fetch_and(volatile atomic_intptr_t __global *object,
+                                 uintptr_t operand);
+intptr_t __ovld atomic_fetch_and(volatile atomic_intptr_t __local *object,
+                                 uintptr_t operand);
+intptr_t __ovld atomic_fetch_min(volatile atomic_intptr_t __global *object,
+                                 uintptr_t opermax);
+intptr_t __ovld atomic_fetch_min(volatile atomic_intptr_t __local *object,
+                                 uintptr_t opermax);
+intptr_t __ovld atomic_fetch_max(volatile atomic_intptr_t __global *object,
+                                 uintptr_t opermax);
+intptr_t __ovld atomic_fetch_max(volatile atomic_intptr_t __local *object,
+                                 uintptr_t opermax);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+
+#ifdef __opencl_c_generic_address_space
+#ifdef __opencl_c_atomic_scope_device
+int __ovld atomic_fetch_add_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order);
+uint __ovld atomic_fetch_add_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_sub_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order);
+uint __ovld atomic_fetch_sub_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_or_explicit(volatile atomic_int *object, int operand,
+                                    memory_order order);
+uint __ovld atomic_fetch_or_explicit(volatile atomic_uint *object, uint operand,
+                                     memory_order order);
+int __ovld atomic_fetch_xor_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order);
+uint __ovld atomic_fetch_xor_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_and_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order);
+uint __ovld atomic_fetch_and_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_min_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order);
+uint __ovld atomic_fetch_min_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_max_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order);
+uint __ovld atomic_fetch_max_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+int __ovld atomic_fetch_add_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order, memory_scope scope);
+uint __ovld atomic_fetch_add_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_sub_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order, memory_scope scope);
+uint __ovld atomic_fetch_sub_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_or_explicit(volatile atomic_int *object, int operand,
+                                    memory_order order, memory_scope scope);
+uint __ovld atomic_fetch_or_explicit(volatile atomic_uint *object, uint operand,
+                                     memory_order order, memory_scope scope);
+int __ovld atomic_fetch_xor_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order, memory_scope scope);
+uint __ovld atomic_fetch_xor_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_and_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order, memory_scope scope);
+uint __ovld atomic_fetch_and_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_min_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order, memory_scope scope);
+uint __ovld atomic_fetch_min_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_max_explicit(volatile atomic_int *object, int operand,
+                                     memory_order order, memory_scope scope);
+uint __ovld atomic_fetch_max_explicit(volatile atomic_uint *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#ifdef __opencl_c_atomic_scope_device
+long __ovld atomic_fetch_add_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_add_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_sub_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_sub_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_or_explicit(volatile atomic_long *object, long operand,
+                                     memory_order order);
+ulong __ovld atomic_fetch_or_explicit(volatile atomic_ulong *object,
+                                      ulong operand, memory_order order);
+long __ovld atomic_fetch_xor_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_xor_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_and_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_and_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_min_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_max_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+long __ovld atomic_fetch_add_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_add_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_sub_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_sub_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_or_explicit(volatile atomic_long *object, long operand,
+                                     memory_order order, memory_scope scope);
+ulong __ovld atomic_fetch_or_explicit(volatile atomic_ulong *object,
+                                      ulong operand, memory_order order,
+                                      memory_scope scope);
+long __ovld atomic_fetch_xor_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_xor_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_and_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_and_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_min_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_max_explicit(volatile atomic_long *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+
+// OpenCL v2.0 s6.13.11.7.5:
+// add/sub: atomic type argument can be uintptr_t/intptr_t, value type argument
+// can be ptrdiff_t. or/xor/and/min/max: atomic type argument can be
+// intptr_t/uintptr_t, value type argument can be intptr_t/uintptr_t.
+
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#ifdef __opencl_c_atomic_scope_device
+uintptr_t __ovld atomic_fetch_add_explicit(volatile atomic_uintptr_t *object,
+                                           ptrdiff_t operand,
+                                           memory_order order);
+uintptr_t __ovld atomic_fetch_sub_explicit(volatile atomic_uintptr_t *object,
+                                           ptrdiff_t operand,
+                                           memory_order order);
+uintptr_t __ovld atomic_fetch_or_explicit(volatile atomic_uintptr_t *object,
+                                          intptr_t operand, memory_order order);
+uintptr_t __ovld atomic_fetch_xor_explicit(volatile atomic_uintptr_t *object,
+                                           intptr_t operand,
+                                           memory_order order);
+uintptr_t __ovld atomic_fetch_and_explicit(volatile atomic_uintptr_t *object,
+                                           intptr_t operand,
+                                           memory_order order);
+uintptr_t __ovld atomic_fetch_min_explicit(volatile atomic_uintptr_t *object,
+                                           intptr_t opermax,
+                                           memory_order minder);
+uintptr_t __ovld atomic_fetch_max_explicit(volatile atomic_uintptr_t *object,
+                                           intptr_t opermax,
+                                           memory_order minder);
+intptr_t __ovld atomic_fetch_or_explicit(volatile atomic_intptr_t *object,
+                                         uintptr_t operand, memory_order order);
+intptr_t __ovld atomic_fetch_xor_explicit(volatile atomic_intptr_t *object,
+                                          uintptr_t operand,
+                                          memory_order order);
+intptr_t __ovld atomic_fetch_and_explicit(volatile atomic_intptr_t *object,
+                                          uintptr_t operand,
+                                          memory_order order);
+intptr_t __ovld atomic_fetch_min_explicit(volatile atomic_intptr_t *object,
+                                          uintptr_t opermax,
+                                          memory_order minder);
+intptr_t __ovld atomic_fetch_max_explicit(volatile atomic_intptr_t *object,
+                                          uintptr_t opermax,
+                                          memory_order minder);
+#endif // __opencl_c_atomic_scope_device
+uintptr_t __ovld atomic_fetch_add_explicit(volatile atomic_uintptr_t *object,
+                                           ptrdiff_t operand,
+                                           memory_order order,
+                                           memory_scope scope);
+uintptr_t __ovld atomic_fetch_sub_explicit(volatile atomic_uintptr_t *object,
+                                           ptrdiff_t operand,
+                                           memory_order order,
+                                           memory_scope scope);
+
+uintptr_t __ovld atomic_fetch_or_explicit(volatile atomic_uintptr_t *object,
+                                          intptr_t operand, memory_order order,
+                                          memory_scope scope);
+uintptr_t __ovld atomic_fetch_xor_explicit(volatile atomic_uintptr_t *object,
+                                           intptr_t operand, memory_order order,
+                                           memory_scope scope);
+uintptr_t __ovld atomic_fetch_and_explicit(volatile atomic_uintptr_t *object,
+                                           intptr_t operand, memory_order order,
+                                           memory_scope scope);
+uintptr_t __ovld atomic_fetch_min_explicit(volatile atomic_uintptr_t *object,
+                                           intptr_t opermax,
+                                           memory_order minder,
+                                           memory_scope scope);
+uintptr_t __ovld atomic_fetch_max_explicit(volatile atomic_uintptr_t *object,
+                                           intptr_t opermax,
+                                           memory_order minder,
+                                           memory_scope scope);
+
+intptr_t __ovld atomic_fetch_or_explicit(volatile atomic_intptr_t *object,
+                                         uintptr_t operand, memory_order order,
+                                         memory_scope scope);
+intptr_t __ovld atomic_fetch_xor_explicit(volatile atomic_intptr_t *object,
+                                          uintptr_t operand, memory_order order,
+                                          memory_scope scope);
+intptr_t __ovld atomic_fetch_and_explicit(volatile atomic_intptr_t *object,
+                                          uintptr_t operand, memory_order order,
+                                          memory_scope scope);
+intptr_t __ovld atomic_fetch_min_explicit(volatile atomic_intptr_t *object,
+                                          uintptr_t opermax,
+                                          memory_order minder,
+                                          memory_scope scope);
+intptr_t __ovld atomic_fetch_max_explicit(volatile atomic_intptr_t *object,
+                                          uintptr_t opermax,
+                                          memory_order minder,
+                                          memory_scope scope);
 #endif
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#ifdef __opencl_c_atomic_scope_device
+int __ovld atomic_fetch_add_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order);
+int __ovld atomic_fetch_add_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order);
+uint __ovld atomic_fetch_add_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order);
+uint __ovld atomic_fetch_add_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_sub_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order);
+int __ovld atomic_fetch_sub_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order);
+uint __ovld atomic_fetch_sub_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order);
+uint __ovld atomic_fetch_sub_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_or_explicit(volatile atomic_int __global *object,
+                                    int operand, memory_order order);
+int __ovld atomic_fetch_or_explicit(volatile atomic_int __local *object,
+                                    int operand, memory_order order);
+uint __ovld atomic_fetch_or_explicit(volatile atomic_uint __global *object,
+                                     uint operand, memory_order order);
+uint __ovld atomic_fetch_or_explicit(volatile atomic_uint __local *object,
+                                     uint operand, memory_order order);
+int __ovld atomic_fetch_xor_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order);
+int __ovld atomic_fetch_xor_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order);
+uint __ovld atomic_fetch_xor_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order);
+uint __ovld atomic_fetch_xor_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_and_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order);
+int __ovld atomic_fetch_and_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order);
+uint __ovld atomic_fetch_and_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order);
+uint __ovld atomic_fetch_and_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_min_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order);
+int __ovld atomic_fetch_min_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order);
+uint __ovld atomic_fetch_min_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order);
+uint __ovld atomic_fetch_min_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order);
+int __ovld atomic_fetch_max_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order);
+int __ovld atomic_fetch_max_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order);
+uint __ovld atomic_fetch_max_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order);
+uint __ovld atomic_fetch_max_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+int __ovld atomic_fetch_add_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+int __ovld atomic_fetch_add_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+uint __ovld atomic_fetch_add_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+uint __ovld atomic_fetch_add_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_sub_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+int __ovld atomic_fetch_sub_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+uint __ovld atomic_fetch_sub_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+uint __ovld atomic_fetch_sub_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_or_explicit(volatile atomic_int __global *object,
+                                    int operand, memory_order order,
+                                    memory_scope scope);
+int __ovld atomic_fetch_or_explicit(volatile atomic_int __local *object,
+                                    int operand, memory_order order,
+                                    memory_scope scope);
+uint __ovld atomic_fetch_or_explicit(volatile atomic_uint __global *object,
+                                     uint operand, memory_order order,
+                                     memory_scope scope);
+uint __ovld atomic_fetch_or_explicit(volatile atomic_uint __local *object,
+                                     uint operand, memory_order order,
+                                     memory_scope scope);
+int __ovld atomic_fetch_xor_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+int __ovld atomic_fetch_xor_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+uint __ovld atomic_fetch_xor_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+uint __ovld atomic_fetch_xor_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_and_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+int __ovld atomic_fetch_and_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+uint __ovld atomic_fetch_and_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+uint __ovld atomic_fetch_and_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_min_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+int __ovld atomic_fetch_min_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+uint __ovld atomic_fetch_min_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+uint __ovld atomic_fetch_min_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+int __ovld atomic_fetch_max_explicit(volatile atomic_int __global *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+int __ovld atomic_fetch_max_explicit(volatile atomic_int __local *object,
+                                     int operand, memory_order order,
+                                     memory_scope scope);
+uint __ovld atomic_fetch_max_explicit(volatile atomic_uint __global *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+uint __ovld atomic_fetch_max_explicit(volatile atomic_uint __local *object,
+                                      uint operand, memory_order order,
+                                      memory_scope scope);
+
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#ifdef __opencl_c_atomic_scope_device
+long __ovld atomic_fetch_add_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order);
+long __ovld atomic_fetch_add_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_add_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order);
+ulong __ovld atomic_fetch_add_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_sub_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order);
+long __ovld atomic_fetch_sub_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_sub_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order);
+ulong __ovld atomic_fetch_sub_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_or_explicit(volatile atomic_long __global *object,
+                                     long operand, memory_order order);
+long __ovld atomic_fetch_or_explicit(volatile atomic_long __local *object,
+                                     long operand, memory_order order);
+ulong __ovld atomic_fetch_or_explicit(volatile atomic_ulong __global *object,
+                                      ulong operand, memory_order order);
+ulong __ovld atomic_fetch_or_explicit(volatile atomic_ulong __local *object,
+                                      ulong operand, memory_order order);
+long __ovld atomic_fetch_xor_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order);
+long __ovld atomic_fetch_xor_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_xor_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order);
+ulong __ovld atomic_fetch_xor_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_and_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order);
+long __ovld atomic_fetch_and_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_and_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order);
+ulong __ovld atomic_fetch_and_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_min_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order);
+long __ovld atomic_fetch_min_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order);
+ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order);
+long __ovld atomic_fetch_max_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order);
+long __ovld atomic_fetch_max_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order);
+ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order);
+ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+long __ovld atomic_fetch_add_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+long __ovld atomic_fetch_add_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_add_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+ulong __ovld atomic_fetch_add_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_sub_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+long __ovld atomic_fetch_sub_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_sub_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+ulong __ovld atomic_fetch_sub_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_or_explicit(volatile atomic_long __global *object,
+                                     long operand, memory_order order,
+                                     memory_scope scope);
+long __ovld atomic_fetch_or_explicit(volatile atomic_long __local *object,
+                                     long operand, memory_order order,
+                                     memory_scope scope);
+ulong __ovld atomic_fetch_or_explicit(volatile atomic_ulong __global *object,
+                                      ulong operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_or_explicit(volatile atomic_ulong __local *object,
+                                      ulong operand, memory_order order,
+                                      memory_scope scope);
+long __ovld atomic_fetch_xor_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+long __ovld atomic_fetch_xor_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_xor_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+ulong __ovld atomic_fetch_xor_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_and_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+long __ovld atomic_fetch_and_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_and_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+ulong __ovld atomic_fetch_and_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_min_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+long __ovld atomic_fetch_min_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+ulong __ovld atomic_fetch_min_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+long __ovld atomic_fetch_max_explicit(volatile atomic_long __global *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+long __ovld atomic_fetch_max_explicit(volatile atomic_long __local *object,
+                                      long operand, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong __global *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+ulong __ovld atomic_fetch_max_explicit(volatile atomic_ulong __local *object,
+                                       ulong operand, memory_order order,
+                                       memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+
+// OpenCL v2.0 s6.13.11.7.5:
+// add/sub: atomic type argument can be uintptr_t/intptr_t, value type argument
+// can be ptrdiff_t. or/xor/and/min/max: atomic type argument can be
+// intptr_t/uintptr_t, value type argument can be intptr_t/uintptr_t.
+
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#ifdef __opencl_c_atomic_scope_device
+uintptr_t __ovld
+atomic_fetch_add_explicit(volatile atomic_uintptr_t __global *object,
+                          ptrdiff_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_add_explicit(volatile atomic_uintptr_t __local *object,
+                          ptrdiff_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_sub_explicit(volatile atomic_uintptr_t __global *object,
+                          ptrdiff_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_sub_explicit(volatile atomic_uintptr_t __local *object,
+                          ptrdiff_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_or_explicit(volatile atomic_uintptr_t __global *object,
+                         intptr_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_or_explicit(volatile atomic_uintptr_t __local *object,
+                         intptr_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_xor_explicit(volatile atomic_uintptr_t __global *object,
+                          intptr_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_xor_explicit(volatile atomic_uintptr_t __local *object,
+                          intptr_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_and_explicit(volatile atomic_uintptr_t __global *object,
+                          intptr_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_and_explicit(volatile atomic_uintptr_t __local *object,
+                          intptr_t operand, memory_order order);
+uintptr_t __ovld
+atomic_fetch_min_explicit(volatile atomic_uintptr_t __global *object,
+                          intptr_t opermax, memory_order minder);
+uintptr_t __ovld
+atomic_fetch_min_explicit(volatile atomic_uintptr_t __local *object,
+                          intptr_t opermax, memory_order minder);
+uintptr_t __ovld
+atomic_fetch_max_explicit(volatile atomic_uintptr_t __global *object,
+                          intptr_t opermax, memory_order minder);
+uintptr_t __ovld
+atomic_fetch_max_explicit(volatile atomic_uintptr_t __local *object,
+                          intptr_t opermax, memory_order minder);
+intptr_t __ovld
+atomic_fetch_or_explicit(volatile atomic_intptr_t __global *object,
+                         uintptr_t operand, memory_order order);
+intptr_t __ovld
+atomic_fetch_or_explicit(volatile atomic_intptr_t __local *object,
+                         uintptr_t operand, memory_order order);
+intptr_t __ovld
+atomic_fetch_xor_explicit(volatile atomic_intptr_t __global *object,
+                          uintptr_t operand, memory_order order);
+intptr_t __ovld
+atomic_fetch_xor_explicit(volatile atomic_intptr_t __local *object,
+                          uintptr_t operand, memory_order order);
+intptr_t __ovld
+atomic_fetch_and_explicit(volatile atomic_intptr_t __global *object,
+                          uintptr_t operand, memory_order order);
+intptr_t __ovld
+atomic_fetch_and_explicit(volatile atomic_intptr_t __local *object,
+                          uintptr_t operand, memory_order order);
+intptr_t __ovld
+atomic_fetch_min_explicit(volatile atomic_intptr_t __global *object,
+                          uintptr_t opermax, memory_order minder);
+intptr_t __ovld
+atomic_fetch_min_explicit(volatile atomic_intptr_t __local *object,
+                          uintptr_t opermax, memory_order minder);
+intptr_t __ovld
+atomic_fetch_max_explicit(volatile atomic_intptr_t __global *object,
+                          uintptr_t opermax, memory_order minder);
+intptr_t __ovld
+atomic_fetch_max_explicit(volatile atomic_intptr_t __local *object,
+                          uintptr_t opermax, memory_order minder);
+#endif // __opencl_c_atomic_scope_device
+uintptr_t __ovld atomic_fetch_add_explicit(
+    volatile atomic_uintptr_t __global *object, ptrdiff_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_add_explicit(
+    volatile atomic_uintptr_t __local *object, ptrdiff_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_sub_explicit(
+    volatile atomic_uintptr_t __global *object, ptrdiff_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_sub_explicit(
+    volatile atomic_uintptr_t __local *object, ptrdiff_t operand,
+    memory_order order, memory_scope scope);
+
+uintptr_t __ovld atomic_fetch_or_explicit(
+    volatile atomic_uintptr_t __global *object, intptr_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_or_explicit(
+    volatile atomic_uintptr_t __local *object, intptr_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_xor_explicit(
+    volatile atomic_uintptr_t __global *object, intptr_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_xor_explicit(
+    volatile atomic_uintptr_t __local *object, intptr_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_and_explicit(
+    volatile atomic_uintptr_t __global *object, intptr_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_and_explicit(
+    volatile atomic_uintptr_t __local *object, intptr_t operand,
+    memory_order order, memory_scope scope);
+uintptr_t __ovld atomic_fetch_min_explicit(
+    volatile atomic_uintptr_t __global *object, intptr_t opermax,
+    memory_order minder, memory_scope scope);
+uintptr_t __ovld atomic_fetch_min_explicit(
+    volatile atomic_uintptr_t __local *object, intptr_t opermax,
+    memory_order minder, memory_scope scope);
+uintptr_t __ovld atomic_fetch_max_explicit(
+    volatile atomic_uintptr_t __global *object, intptr_t opermax,
+    memory_order minder, memory_scope scope);
+uintptr_t __ovld atomic_fetch_max_explicit(
+    volatile atomic_uintptr_t __local *object, intptr_t opermax,
+    memory_order minder, memory_scope scope);
+
+intptr_t __ovld atomic_fetch_or_explicit(
+    volatile atomic_intptr_t __global *object, uintptr_t operand,
+    memory_order order, memory_scope scope);
+intptr_t __ovld atomic_fetch_or_explicit(
+    volatile atomic_intptr_t __local *object, uintptr_t operand,
+    memory_order order, memory_scope scope);
+intptr_t __ovld atomic_fetch_xor_explicit(
+    volatile atomic_intptr_t __global *object, uintptr_t operand,
+    memory_order order, memory_scope scope);
+intptr_t __ovld atomic_fetch_xor_explicit(
+    volatile atomic_intptr_t __local *object, uintptr_t operand,
+    memory_order order, memory_scope scope);
+intptr_t __ovld atomic_fetch_and_explicit(
+    volatile atomic_intptr_t __global *object, uintptr_t operand,
+    memory_order order, memory_scope scope);
+intptr_t __ovld atomic_fetch_and_explicit(
+    volatile atomic_intptr_t __local *object, uintptr_t operand,
+    memory_order order, memory_scope scope);
+intptr_t __ovld atomic_fetch_min_explicit(
+    volatile atomic_intptr_t __global *object, uintptr_t opermax,
+    memory_order minder, memory_scope scope);
+intptr_t __ovld atomic_fetch_min_explicit(
+    volatile atomic_intptr_t __local *object, uintptr_t opermax,
+    memory_order minder, memory_scope scope);
+intptr_t __ovld atomic_fetch_max_explicit(
+    volatile atomic_intptr_t __global *object, uintptr_t opermax,
+    memory_order minder, memory_scope scope);
+intptr_t __ovld atomic_fetch_max_explicit(
+    volatile atomic_intptr_t __local *object, uintptr_t opermax,
+    memory_order minder, memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
 
 // atomic_store()
 
+#if defined(__opencl_c_atomic_scope_device) &&                                 \
+    defined(__opencl_c_atomic_order_seq_cst)
+
+#ifdef __opencl_c_generic_address_space
 void __ovld atomic_store(volatile atomic_int *object, int desired);
-void __ovld atomic_store_explicit(volatile atomic_int *object, int desired, memory_order order);
-void __ovld atomic_store_explicit(volatile atomic_int *object, int desired, memory_order order, memory_scope scope);
 void __ovld atomic_store(volatile atomic_uint *object, uint desired);
-void __ovld atomic_store_explicit(volatile atomic_uint *object, uint desired, memory_order order);
-void __ovld atomic_store_explicit(volatile atomic_uint *object, uint desired, memory_order order, memory_scope scope);
 void __ovld atomic_store(volatile atomic_float *object, float desired);
-void __ovld atomic_store_explicit(volatile atomic_float *object, float desired, memory_order order);
-void __ovld atomic_store_explicit(volatile atomic_float *object, float desired, memory_order order, memory_scope scope);
 #if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 void __ovld atomic_store(volatile atomic_double *object, double desired);
-void __ovld atomic_store_explicit(volatile atomic_double *object, double desired, memory_order order);
-void __ovld atomic_store_explicit(volatile atomic_double *object, double desired, memory_order order, memory_scope scope);
-#endif //cl_khr_fp64
+#endif
 void __ovld atomic_store(volatile atomic_long *object, long desired);
-void __ovld atomic_store_explicit(volatile atomic_long *object, long desired, memory_order order);
-void __ovld atomic_store_explicit(volatile atomic_long *object, long desired, memory_order order, memory_scope scope);
 void __ovld atomic_store(volatile atomic_ulong *object, ulong desired);
-void __ovld atomic_store_explicit(volatile atomic_ulong *object, ulong desired, memory_order order);
-void __ovld atomic_store_explicit(volatile atomic_ulong *object, ulong desired, memory_order order, memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+void __ovld atomic_store(volatile atomic_int __global *object, int desired);
+void __ovld atomic_store(volatile atomic_int __local *object, int desired);
+void __ovld atomic_store(volatile atomic_uint __global *object, uint desired);
+void __ovld atomic_store(volatile atomic_uint __local *object, uint desired);
+void __ovld atomic_store(volatile atomic_float __global *object, float desired);
+void __ovld atomic_store(volatile atomic_float __local *object, float desired);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+void __ovld atomic_store(volatile atomic_double __global *object,
+                         double desired);
+void __ovld atomic_store(volatile atomic_double __local *object,
+                         double desired);
 #endif
+void __ovld atomic_store(volatile atomic_long __global *object, long desired);
+void __ovld atomic_store(volatile atomic_long __local *object, long desired);
+void __ovld atomic_store(volatile atomic_ulong __global *object, ulong desired);
+void __ovld atomic_store(volatile atomic_ulong __local *object, ulong desired);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+
+#ifdef __opencl_c_generic_address_space
+#ifdef __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_int *object, int desired,
+                                  memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_uint *object, uint desired,
+                                  memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_float *object, float desired,
+                                  memory_order order);
+#endif // __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_int *object, int desired,
+                                  memory_order order, memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_uint *object, uint desired,
+                                  memory_order order, memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_float *object, float desired,
+                                  memory_order order, memory_scope scope);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_double *object,
+                                  double desired, memory_order order);
+#endif //__opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_double *object,
+                                  double desired, memory_order order,
+                                  memory_scope scope);
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_long *object, long desired,
+                                  memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_ulong *object, ulong desired,
+                                  memory_order order);
+#endif //__opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_long *object, long desired,
+                                  memory_order order, memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_ulong *object, ulong desired,
+                                  memory_order order, memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#ifdef __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_int __global *object,
+                                  int desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_int __local *object,
+                                  int desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_uint __global *object,
+                                  uint desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_uint __local *object,
+                                  uint desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_float __global *object,
+                                  float desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_float __local *object,
+                                  float desired, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_int __global *object,
+                                  int desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_int __local *object,
+                                  int desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_uint __global *object,
+                                  uint desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_uint __local *object,
+                                  uint desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_float __global *object,
+                                  float desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_float __local *object,
+                                  float desired, memory_order order,
+                                  memory_scope scope);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_double __global *object,
+                                  double desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_double __local *object,
+                                  double desired, memory_order order);
+#endif //__opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_double __global *object,
+                                  double desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_double __local *object,
+                                  double desired, memory_order order,
+                                  memory_scope scope);
+#endif // cl_khr_fp64
+#ifdef __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_long __global *object,
+                                  long desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_long __local *object,
+                                  long desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_ulong __global *object,
+                                  ulong desired, memory_order order);
+void __ovld atomic_store_explicit(volatile atomic_ulong __local *object,
+                                  ulong desired, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+void __ovld atomic_store_explicit(volatile atomic_long __global *object,
+                                  long desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_long __local *object,
+                                  long desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_ulong __global *object,
+                                  ulong desired, memory_order order,
+                                  memory_scope scope);
+void __ovld atomic_store_explicit(volatile atomic_ulong __local *object,
+                                  ulong desired, memory_order order,
+                                  memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
 
 // atomic_load()
-
+#ifdef __opencl_c_generic_address_space
+#if defined(__opencl_c_atomic_scope_device) &&                                 \
+    defined(__opencl_c_atomic_order_seq_cst)
 int __ovld atomic_load(volatile atomic_int *object);
-int __ovld atomic_load_explicit(volatile atomic_int *object, memory_order order);
-int __ovld atomic_load_explicit(volatile atomic_int *object, memory_order order, memory_scope scope);
 uint __ovld atomic_load(volatile atomic_uint *object);
-uint __ovld atomic_load_explicit(volatile atomic_uint *object, memory_order order);
-uint __ovld atomic_load_explicit(volatile atomic_uint *object, memory_order order, memory_scope scope);
 float __ovld atomic_load(volatile atomic_float *object);
-float __ovld atomic_load_explicit(volatile atomic_float *object, memory_order order);
-float __ovld atomic_load_explicit(volatile atomic_float *object, memory_order order, memory_scope scope);
 #if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld atomic_load(volatile atomic_double *object);
-double __ovld atomic_load_explicit(volatile atomic_double *object, memory_order order);
-double __ovld atomic_load_explicit(volatile atomic_double *object, memory_order order, memory_scope scope);
-#endif //cl_khr_fp64
+#endif // cl_khr_fp64
 long __ovld atomic_load(volatile atomic_long *object);
-long __ovld atomic_load_explicit(volatile atomic_long *object, memory_order order);
-long __ovld atomic_load_explicit(volatile atomic_long *object, memory_order order, memory_scope scope);
 ulong __ovld atomic_load(volatile atomic_ulong *object);
-ulong __ovld atomic_load_explicit(volatile atomic_ulong *object, memory_order order);
-ulong __ovld atomic_load_explicit(volatile atomic_ulong *object, memory_order order, memory_scope scope);
-#endif
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+#endif //__opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#if defined(__opencl_c_atomic_scope_device) &&                                 \
+    defined(__opencl_c_atomic_order_seq_cst)
+int __ovld atomic_load(volatile atomic_int __global *object);
+int __ovld atomic_load(volatile atomic_int __local *object);
+uint __ovld atomic_load(volatile atomic_uint __global *object);
+uint __ovld atomic_load(volatile atomic_uint __local *object);
+float __ovld atomic_load(volatile atomic_float __global *object);
+float __ovld atomic_load(volatile atomic_float __local *object);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+double __ovld atomic_load(volatile atomic_double __global *object);
+double __ovld atomic_load(volatile atomic_double __local *object);
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+long __ovld atomic_load(volatile atomic_long __global *object);
+long __ovld atomic_load(volatile atomic_long __local *object);
+ulong __ovld atomic_load(volatile atomic_ulong __global *object);
+ulong __ovld atomic_load(volatile atomic_ulong __local *object);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+
+#ifdef __opencl_c_generic_address_space
+#ifdef __opencl_c_atomic_scope_device
+int __ovld atomic_load_explicit(volatile atomic_int *object,
+                                memory_order order);
+uint __ovld atomic_load_explicit(volatile atomic_uint *object,
+                                 memory_order order);
+float __ovld atomic_load_explicit(volatile atomic_float *object,
+                                  memory_order order);
+#endif // __opencl_c_atomic_scope_device
+int __ovld atomic_load_explicit(volatile atomic_int *object, memory_order order,
+                                memory_scope scope);
+uint __ovld atomic_load_explicit(volatile atomic_uint *object,
+                                 memory_order order, memory_scope scope);
+float __ovld atomic_load_explicit(volatile atomic_float *object,
+                                  memory_order order, memory_scope scope);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+double __ovld atomic_load_explicit(volatile atomic_double *object,
+                                   memory_order order);
+#endif // __opencl_c_atomic_scope_device
+double __ovld atomic_load_explicit(volatile atomic_double *object,
+                                   memory_order order, memory_scope scope);
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+long __ovld atomic_load_explicit(volatile atomic_long *object,
+                                 memory_order order);
+ulong __ovld atomic_load_explicit(volatile atomic_ulong *object,
+                                  memory_order order);
+#endif //__opencl_c_atomic_scope_device
+long __ovld atomic_load_explicit(volatile atomic_long *object,
+                                 memory_order order, memory_scope scope);
+ulong __ovld atomic_load_explicit(volatile atomic_ulong *object,
+                                  memory_order order, memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#ifdef __opencl_c_atomic_scope_device
+int __ovld atomic_load_explicit(volatile atomic_int __global *object,
+                                memory_order order);
+int __ovld atomic_load_explicit(volatile atomic_int __local *object,
+                                memory_order order);
+uint __ovld atomic_load_explicit(volatile atomic_uint __global *object,
+                                 memory_order order);
+uint __ovld atomic_load_explicit(volatile atomic_uint __local *object,
+                                 memory_order order);
+float __ovld atomic_load_explicit(volatile atomic_float __global *object,
+                                  memory_order order);
+float __ovld atomic_load_explicit(volatile atomic_float __local *object,
+                                  memory_order order);
+#endif // __opencl_c_atomic_scope_device
+int __ovld atomic_load_explicit(volatile atomic_int __global *object,
+                                memory_order order, memory_scope scope);
+int __ovld atomic_load_explicit(volatile atomic_int __local *object,
+                                memory_order order, memory_scope scope);
+uint __ovld atomic_load_explicit(volatile atomic_uint __global *object,
+                                 memory_order order, memory_scope scope);
+uint __ovld atomic_load_explicit(volatile atomic_uint __local *object,
+                                 memory_order order, memory_scope scope);
+float __ovld atomic_load_explicit(volatile atomic_float __global *object,
+                                  memory_order order, memory_scope scope);
+float __ovld atomic_load_explicit(volatile atomic_float __local *object,
+                                  memory_order order, memory_scope scope);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+double __ovld atomic_load_explicit(volatile atomic_double __global *object,
+                                   memory_order order);
+double __ovld atomic_load_explicit(volatile atomic_double __local *object,
+                                   memory_order order);
+#endif // __opencl_c_atomic_scope_device
+double __ovld atomic_load_explicit(volatile atomic_double __global *object,
+                                   memory_order order, memory_scope scope);
+double __ovld atomic_load_explicit(volatile atomic_double __local *object,
+                                   memory_order order, memory_scope scope);
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+long __ovld atomic_load_explicit(volatile atomic_long __global *object,
+                                 memory_order order);
+long __ovld atomic_load_explicit(volatile atomic_long __local *object,
+                                 memory_order order);
+ulong __ovld atomic_load_explicit(volatile atomic_ulong __global *object,
+                                  memory_order order);
+ulong __ovld atomic_load_explicit(volatile atomic_ulong __local *object,
+                                  memory_order order);
+#endif // __opencl_c_atomic_scope_device
+long __ovld atomic_load_explicit(volatile atomic_long __global *object,
+                                 memory_order order, memory_scope scope);
+long __ovld atomic_load_explicit(volatile atomic_long __local *object,
+                                 memory_order order, memory_scope scope);
+ulong __ovld atomic_load_explicit(volatile atomic_ulong __global *object,
+                                  memory_order order, memory_scope scope);
+ulong __ovld atomic_load_explicit(volatile atomic_ulong __local *object,
+                                  memory_order order, memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
 
 // atomic_exchange()
-
+#ifdef __opencl_c_generic_address_space
+#if defined(__opencl_c_atomic_scope_device) &&                                 \
+    defined(__opencl_c_atomic_order_seq_cst)
 int __ovld atomic_exchange(volatile atomic_int *object, int desired);
-int __ovld atomic_exchange_explicit(volatile atomic_int *object, int desired, memory_order order);
-int __ovld atomic_exchange_explicit(volatile atomic_int *object, int desired, memory_order order, memory_scope scope);
 uint __ovld atomic_exchange(volatile atomic_uint *object, uint desired);
-uint __ovld atomic_exchange_explicit(volatile atomic_uint *object, uint desired, memory_order order);
-uint __ovld atomic_exchange_explicit(volatile atomic_uint *object, uint desired, memory_order order, memory_scope scope);
 float __ovld atomic_exchange(volatile atomic_float *object, float desired);
-float __ovld atomic_exchange_explicit(volatile atomic_float *object, float desired, memory_order order);
-float __ovld atomic_exchange_explicit(volatile atomic_float *object, float desired, memory_order order, memory_scope scope);
 #if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld atomic_exchange(volatile atomic_double *object, double desired);
-double __ovld atomic_exchange_explicit(volatile atomic_double *object, double desired, memory_order order);
-double __ovld atomic_exchange_explicit(volatile atomic_double *object, double desired, memory_order order, memory_scope scope);
-#endif //cl_khr_fp64
+#endif
 long __ovld atomic_exchange(volatile atomic_long *object, long desired);
-long __ovld atomic_exchange_explicit(volatile atomic_long *object, long desired, memory_order order);
-long __ovld atomic_exchange_explicit(volatile atomic_long *object, long desired, memory_order order, memory_scope scope);
 ulong __ovld atomic_exchange(volatile atomic_ulong *object, ulong desired);
-ulong __ovld atomic_exchange_explicit(volatile atomic_ulong *object, ulong desired, memory_order order);
-ulong __ovld atomic_exchange_explicit(volatile atomic_ulong *object, ulong desired, memory_order order, memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#if defined(__opencl_c_atomic_scope_device) &&                                 \
+    defined(__opencl_c_atomic_order_seq_cst)
+int __ovld atomic_exchange(volatile atomic_int __global *object, int desired);
+int __ovld atomic_exchange(volatile atomic_int __local *object, int desired);
+uint __ovld atomic_exchange(volatile atomic_uint __global *object,
+                            uint desired);
+uint __ovld atomic_exchange(volatile atomic_uint __local *object, uint desired);
+float __ovld atomic_exchange(volatile atomic_float __global *object,
+                             float desired);
+float __ovld atomic_exchange(volatile atomic_float __local *object,
+                             float desired);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+double __ovld atomic_exchange(volatile atomic_double __global *object,
+                              double desired);
+double __ovld atomic_exchange(volatile atomic_double __local *object,
+                              double desired);
 #endif
+long __ovld atomic_exchange(volatile atomic_long __global *object,
+                            long desired);
+long __ovld atomic_exchange(volatile atomic_long __local *object, long desired);
+ulong __ovld atomic_exchange(volatile atomic_ulong __global *object,
+                             ulong desired);
+ulong __ovld atomic_exchange(volatile atomic_ulong __local *object,
+                             ulong desired);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+
+#ifdef __opencl_c_generic_address_space
+#ifdef __opencl_c_atomic_scope_device
+int __ovld atomic_exchange_explicit(volatile atomic_int *object, int desired,
+                                    memory_order order);
+uint __ovld atomic_exchange_explicit(volatile atomic_uint *object, uint desired,
+                                     memory_order order);
+float __ovld atomic_exchange_explicit(volatile atomic_float *object,
+                                      float desired, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+int __ovld atomic_exchange_explicit(volatile atomic_int *object, int desired,
+                                    memory_order order, memory_scope scope);
+uint __ovld atomic_exchange_explicit(volatile atomic_uint *object, uint desired,
+                                     memory_order order, memory_scope scope);
+float __ovld atomic_exchange_explicit(volatile atomic_float *object,
+                                      float desired, memory_order order,
+                                      memory_scope scope);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+double __ovld atomic_exchange_explicit(volatile atomic_double *object,
+                                       double desired, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+double __ovld atomic_exchange_explicit(volatile atomic_double *object,
+                                       double desired, memory_order order,
+                                       memory_scope scope);
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+long __ovld atomic_exchange_explicit(volatile atomic_long *object, long desired,
+                                     memory_order order);
+ulong __ovld atomic_exchange_explicit(volatile atomic_ulong *object,
+                                      ulong desired, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+long __ovld atomic_exchange_explicit(volatile atomic_long *object, long desired,
+                                     memory_order order, memory_scope scope);
+ulong __ovld atomic_exchange_explicit(volatile atomic_ulong *object,
+                                      ulong desired, memory_order order,
+                                      memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#ifdef __opencl_c_atomic_scope_device
+int __ovld atomic_exchange_explicit(volatile atomic_int __global *object,
+                                    int desired, memory_order order);
+int __ovld atomic_exchange_explicit(volatile atomic_int __local *object,
+                                    int desired, memory_order order);
+uint __ovld atomic_exchange_explicit(volatile atomic_uint __global *object,
+                                     uint desired, memory_order order);
+uint __ovld atomic_exchange_explicit(volatile atomic_uint __local *object,
+                                     uint desired, memory_order order);
+float __ovld atomic_exchange_explicit(volatile atomic_float __global *object,
+                                      float desired, memory_order order);
+float __ovld atomic_exchange_explicit(volatile atomic_float __local *object,
+                                      float desired, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+int __ovld atomic_exchange_explicit(volatile atomic_int __global *object,
+                                    int desired, memory_order order,
+                                    memory_scope scope);
+int __ovld atomic_exchange_explicit(volatile atomic_int __local *object,
+                                    int desired, memory_order order,
+                                    memory_scope scope);
+uint __ovld atomic_exchange_explicit(volatile atomic_uint __global *object,
+                                     uint desired, memory_order order,
+                                     memory_scope scope);
+uint __ovld atomic_exchange_explicit(volatile atomic_uint __local *object,
+                                     uint desired, memory_order order,
+                                     memory_scope scope);
+float __ovld atomic_exchange_explicit(volatile atomic_float __global *object,
+                                      float desired, memory_order order,
+                                      memory_scope scope);
+float __ovld atomic_exchange_explicit(volatile atomic_float __local *object,
+                                      float desired, memory_order order,
+                                      memory_scope scope);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+double __ovld atomic_exchange_explicit(volatile atomic_double __global *object,
+                                       double desired, memory_order order);
+double __ovld atomic_exchange_explicit(volatile atomic_double __local *object,
+                                       double desired, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+double __ovld atomic_exchange_explicit(volatile atomic_double __global *object,
+                                       double desired, memory_order order,
+                                       memory_scope scope);
+double __ovld atomic_exchange_explicit(volatile atomic_double __local *object,
+                                       double desired, memory_order order,
+                                       memory_scope scope);
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+long __ovld atomic_exchange_explicit(volatile atomic_long __global *object,
+                                     long desired, memory_order order);
+long __ovld atomic_exchange_explicit(volatile atomic_long __local *object,
+                                     long desired, memory_order order);
+ulong __ovld atomic_exchange_explicit(volatile atomic_ulong __global *object,
+                                      ulong desired, memory_order order);
+ulong __ovld atomic_exchange_explicit(volatile atomic_ulong __local *object,
+                                      ulong desired, memory_order order);
+#endif // __opencl_c_atomic_scope_device
+long __ovld atomic_exchange_explicit(volatile atomic_long __global *object,
+                                     long desired, memory_order order,
+                                     memory_scope scope);
+long __ovld atomic_exchange_explicit(volatile atomic_long __local *object,
+                                     long desired, memory_order order,
+                                     memory_scope scope);
+ulong __ovld atomic_exchange_explicit(volatile atomic_ulong __global *object,
+                                      ulong desired, memory_order order,
+                                      memory_scope scope);
+ulong __ovld atomic_exchange_explicit(volatile atomic_ulong __local *object,
+                                      ulong desired, memory_order order,
+                                      memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
 
 // atomic_compare_exchange_strong() and atomic_compare_exchange_weak()
-
-bool __ovld atomic_compare_exchange_strong(volatile atomic_int *object, int *expected, int desired);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_int *object, int *expected,
-                                                                                 int desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_int *object, int *expected,
-                                                                                 int desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_strong(volatile atomic_uint *object, uint *expected, uint desired);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_uint *object, uint *expected,
-                                                                                 uint desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_uint *object, uint *expected,
-                                                                                 uint desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_weak(volatile atomic_int *object, int *expected, int desired);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_int *object, int *expected,
-                                                                                 int desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_int *object, int *expected,
-                                                                                 int desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_weak(volatile atomic_uint *object, uint *expected, uint desired);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_uint *object, uint *expected,
-                                                                                 uint desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_uint *object, uint *expected,
-                                                                                 uint desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_strong(volatile atomic_float *object, float *expected, float desired);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_float *object, float *expected,
-                                                                                 float desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_float *object, float *expected,
-                                                                                 float desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_weak(volatile atomic_float *object, float *expected, float desired);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_float *object, float *expected,
-                                                                                 float desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_float *object, float *expected,
-                                                                                 float desired, memory_order success, memory_order failure, memory_scope scope);
+#ifdef __opencl_c_generic_address_space
+#if defined(__opencl_c_atomic_scope_device) &&                                 \
+    defined(__opencl_c_atomic_order_seq_cst)
+bool __ovld atomic_compare_exchange_strong(volatile atomic_int *object,
+                                           int *expected, int desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_uint *object,
+                                           uint *expected, uint desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_int *object,
+                                         int *expected, int desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_uint *object,
+                                         uint *expected, uint desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_float *object,
+                                           float *expected, float desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_float *object,
+                                         float *expected, float desired);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+bool __ovld atomic_compare_exchange_strong(volatile atomic_double *object,
+                                           double *expected, double desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_double *object,
+                                         double *expected, double desired);
+#endif
+bool __ovld atomic_compare_exchange_strong(volatile atomic_long *object,
+                                           long *expected, long desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_long *object,
+                                         long *expected, long desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_ulong *object,
+                                           ulong *expected, ulong desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_ulong *object,
+                                         ulong *expected, ulong desired);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+#endif //  __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#if defined(__opencl_c_atomic_scope_device) &&                                 \
+    defined(__opencl_c_atomic_order_seq_cst)
+bool __ovld atomic_compare_exchange_strong(volatile atomic_int __global *object,
+                                           int __global *expected, int desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_int __local *object,
+                                           int __global *expected, int desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_int __global *object,
+                                           int __local *expected, int desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_int __local *object,
+                                           int __local *expected, int desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_int __global *object,
+                                           int __private *expected,
+                                           int desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_int __local *object,
+                                           int __private *expected,
+                                           int desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_uint __global *object,
+                               uint __global *expected, uint desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_uint __local *object,
+                                           uint __global *expected,
+                                           uint desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_uint __global *object,
+                               uint __local *expected, uint desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_uint __local *object,
+                                           uint __local *expected,
+                                           uint desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_uint __global *object,
+                               uint __private *expected, uint desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_uint __local *object,
+                                           uint __private *expected,
+                                           uint desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_int __global *object,
+                                         int __global *expected, int desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_int __local *object,
+                                         int __global *expected, int desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_int __global *object,
+                                         int __local *expected, int desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_int __local *object,
+                                         int __local *expected, int desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_int __global *object,
+                                         int __private *expected, int desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_int __local *object,
+                                         int __private *expected, int desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_uint __global *object,
+                                         uint __global *expected, uint desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_uint __local *object,
+                                         uint __global *expected, uint desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_uint __global *object,
+                                         uint __local *expected, uint desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_uint __local *object,
+                                         uint __local *expected, uint desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_uint __global *object,
+                                         uint __private *expected,
+                                         uint desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_uint __local *object,
+                                         uint __private *expected,
+                                         uint desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_float __global *object,
+                               float __global *expected, float desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_float __local *object,
+                               float __global *expected, float desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_float __global *object,
+                               float __local *expected, float desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_float __local *object,
+                               float __local *expected, float desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_float __global *object,
+                               float __private *expected, float desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_float __local *object,
+                               float __private *expected, float desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_float __global *object,
+                                         float __global *expected,
+                                         float desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_float __local *object,
+                                         float __global *expected,
+                                         float desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_float __global *object,
+                                         float __local *expected,
+                                         float desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_float __local *object,
+                                         float __local *expected,
+                                         float desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_float __global *object,
+                                         float __private *expected,
+                                         float desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_float __local *object,
+                                         float __private *expected,
+                                         float desired);
 #if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
-#ifdef cl_khr_fp64
-bool __ovld atomic_compare_exchange_strong(volatile atomic_double *object, double *expected, double desired);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_double *object, double *expected,
-                                                                                 double desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_double *object, double *expected,
-                                                                                 double desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_weak(volatile atomic_double *object, double *expected, double desired);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_double *object, double *expected,
-                                                                                 double desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_double *object, double *expected,
-                                                                                 double desired, memory_order success, memory_order failure, memory_scope scope);
-#endif //cl_khr_fp64
-bool __ovld atomic_compare_exchange_strong(volatile atomic_long *object, long *expected, long desired);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_long *object, long *expected,
-                                                                                 long desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_long *object, long *expected,
-                                                                                 long desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_weak(volatile atomic_long *object, long *expected, long desired);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_long *object, long *expected,
-                                                                                 long desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_long *object, long *expected,
-                                                                                 long desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_strong(volatile atomic_ulong *object, ulong *expected, ulong desired);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_ulong *object, ulong *expected,
-                                                                                 ulong desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_ulong *object, ulong *expected,
-                                                                                 ulong desired, memory_order success, memory_order failure, memory_scope scope);
-bool __ovld atomic_compare_exchange_weak(volatile atomic_ulong *object, ulong *expected, ulong desired);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_ulong *object, ulong *expected,
-                                                                                 ulong desired, memory_order success, memory_order failure);
-bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_ulong *object, ulong *expected,
-                                                                                 ulong desired, memory_order success, memory_order failure, memory_scope scope);
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_double __global *object,
+                               double __global *expected, double desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_double __local *object,
+                               double __global *expected, double desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_double __global *object,
+                               double __local *expected, double desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_double __local *object,
+                               double __local *expected, double desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_double __global *object,
+                               double __private *expected, double desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_double __local *object,
+                               double __private *expected, double desired);
+bool __ovld
+atomic_compare_exchange_weak(volatile atomic_double __global *object,
+                             double __global *expected, double desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_double __local *object,
+                                         double __global *expected,
+                                         double desired);
+bool __ovld
+atomic_compare_exchange_weak(volatile atomic_double __global *object,
+                             double __local *expected, double desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_double __local *object,
+                                         double __local *expected,
+                                         double desired);
+bool __ovld
+atomic_compare_exchange_weak(volatile atomic_double __global *object,
+                             double __private *expected, double desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_double __local *object,
+                                         double __private *expected,
+                                         double desired);
 #endif
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_long __global *object,
+                               long __global *expected, long desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_long __local *object,
+                                           long __global *expected,
+                                           long desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_long __global *object,
+                               long __local *expected, long desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_long __local *object,
+                                           long __local *expected,
+                                           long desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_long __global *object,
+                               long __private *expected, long desired);
+bool __ovld atomic_compare_exchange_strong(volatile atomic_long __local *object,
+                                           long __private *expected,
+                                           long desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_long __global *object,
+                                         long __global *expected, long desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_long __local *object,
+                                         long __global *expected, long desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_long __global *object,
+                                         long __local *expected, long desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_long __local *object,
+                                         long __local *expected, long desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_long __global *object,
+                                         long __private *expected,
+                                         long desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_long __local *object,
+                                         long __private *expected,
+                                         long desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_ulong __global *object,
+                               ulong __global *expected, ulong desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_ulong __local *object,
+                               ulong __global *expected, ulong desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_ulong __global *object,
+                               ulong __local *expected, ulong desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_ulong __local *object,
+                               ulong __local *expected, ulong desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_ulong __global *object,
+                               ulong __private *expected, ulong desired);
+bool __ovld
+atomic_compare_exchange_strong(volatile atomic_ulong __local *object,
+                               ulong __private *expected, ulong desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_ulong __global *object,
+                                         ulong __global *expected,
+                                         ulong desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_ulong __local *object,
+                                         ulong __global *expected,
+                                         ulong desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_ulong __global *object,
+                                         ulong __local *expected,
+                                         ulong desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_ulong __local *object,
+                                         ulong __local *expected,
+                                         ulong desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_ulong __global *object,
+                                         ulong __private *expected,
+                                         ulong desired);
+bool __ovld atomic_compare_exchange_weak(volatile atomic_ulong __local *object,
+                                         ulong __private *expected,
+                                         ulong desired);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+
+#ifdef __opencl_c_generic_address_space
+#ifdef __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_int *object,
+                                                    int *expected, int desired,
+                                                    memory_order success,
+                                                    memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint *object, uint *expected, uint desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_int *object,
+                                                  int *expected, int desired,
+                                                  memory_order success,
+                                                  memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_uint *object,
+                                                  uint *expected, uint desired,
+                                                  memory_order success,
+                                                  memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float *object, float *expected, float desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_float *object,
+                                                  float *expected,
+                                                  float desired,
+                                                  memory_order success,
+                                                  memory_order failure);
+#endif // __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(volatile atomic_int *object,
+                                                    int *expected, int desired,
+                                                    memory_order success,
+                                                    memory_order failure,
+                                                    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint *object, uint *expected, uint desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_int *object,
+                                                  int *expected, int desired,
+                                                  memory_order success,
+                                                  memory_order failure,
+                                                  memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_uint *object,
+                                                  uint *expected, uint desired,
+                                                  memory_order success,
+                                                  memory_order failure,
+                                                  memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float *object, float *expected, float desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float *object, float *expected, float desired,
+    memory_order success, memory_order failure, memory_scope scope);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double *object, double *expected, double desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double *object, double *expected, double desired,
+    memory_order success, memory_order failure);
+#endif // __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double *object, double *expected, double desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double *object, double *expected, double desired,
+    memory_order success, memory_order failure, memory_scope scope);
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long *object, long *expected, long desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_long *object,
+                                                  long *expected, long desired,
+                                                  memory_order success,
+                                                  memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong *object, ulong *expected, ulong desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_ulong *object,
+                                                  ulong *expected,
+                                                  ulong desired,
+                                                  memory_order success,
+                                                  memory_order failure);
+#endif // __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long *object, long *expected, long desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(volatile atomic_long *object,
+                                                  long *expected, long desired,
+                                                  memory_order success,
+                                                  memory_order failure,
+                                                  memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong *object, ulong *expected, ulong desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong *object, ulong *expected, ulong desired,
+    memory_order success, memory_order failure, memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#ifdef __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __global *object, int __global *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __local *object, int __global *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __global *object, int __local *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __local *object, int __local *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __global *object, int __private *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __local *object, int __private *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __global *object, uint __global *expected,
+    uint desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __local *object, uint __global *expected, uint desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __global *object, uint __local *expected, uint desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __local *object, uint __local *expected, uint desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __global *object, uint __private *expected,
+    uint desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __local *object, uint __private *expected,
+    uint desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __global *object, int __global *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __local *object, int __global *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __global *object, int __local *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __local *object, int __local *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __global *object, int __private *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __local *object, int __private *expected, int desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __global *object, uint __global *expected,
+    uint desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __local *object, uint __global *expected, uint desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __global *object, uint __local *expected, uint desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __local *object, uint __local *expected, uint desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __global *object, uint __private *expected,
+    uint desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __local *object, uint __private *expected,
+    uint desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __global *object, float __global *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __local *object, float __global *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __global *object, float __local *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __local *object, float __local *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __global *object, float __private *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __local *object, float __private *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __global *object, float __global *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __local *object, float __global *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __global *object, float __local *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __local *object, float __local *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __global *object, float __private *expected,
+    float desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __local *object, float __private *expected,
+    float desired, memory_order success, memory_order failure);
+#endif // __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __global *object, int __global *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __local *object, int __global *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __global *object, int __local *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __local *object, int __local *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __global *object, int __private *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_int __local *object, int __private *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __global *object, uint __global *expected,
+    uint desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __local *object, uint __global *expected, uint desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __global *object, uint __local *expected, uint desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __local *object, uint __local *expected, uint desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __global *object, uint __private *expected,
+    uint desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_uint __local *object, uint __private *expected,
+    uint desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __global *object, int __global *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __local *object, int __global *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __global *object, int __local *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __local *object, int __local *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __global *object, int __private *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_int __local *object, int __private *expected, int desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __global *object, uint __global *expected,
+    uint desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __local *object, uint __global *expected, uint desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __global *object, uint __local *expected, uint desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __local *object, uint __local *expected, uint desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __global *object, uint __private *expected,
+    uint desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_uint __local *object, uint __private *expected,
+    uint desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __global *object, float __global *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __local *object, float __global *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __global *object, float __local *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __local *object, float __local *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __global *object, float __private *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_float __local *object, float __private *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __global *object, float __global *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __local *object, float __global *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __global *object, float __local *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __local *object, float __local *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __global *object, float __private *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_float __local *object, float __private *expected,
+    float desired, memory_order success, memory_order failure,
+    memory_scope scope);
+#if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __global *object, double __global *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __local *object, double __global *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __global *object, double __local *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __local *object, double __local *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __global *object, double __private *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __local *object, double __private *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __global *object, double __global *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __local *object, double __global *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __global *object, double __local *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __local *object, double __local *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __global *object, double __private *expected,
+    double desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __local *object, double __private *expected,
+    double desired, memory_order success, memory_order failure);
+#endif // __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __global *object, double __global *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __local *object, double __global *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __global *object, double __local *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __local *object, double __local *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __global *object, double __private *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_double __local *object, double __private *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __global *object, double __global *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __local *object, double __global *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __global *object, double __local *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __local *object, double __local *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __global *object, double __private *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_double __local *object, double __private *expected,
+    double desired, memory_order success, memory_order failure,
+    memory_scope scope);
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#ifdef __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __global *object, long __global *expected,
+    long desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __local *object, long __global *expected, long desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __global *object, long __local *expected, long desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __local *object, long __local *expected, long desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __global *object, long __private *expected,
+    long desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __local *object, long __private *expected,
+    long desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __global *object, long __global *expected,
+    long desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __local *object, long __global *expected, long desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __global *object, long __local *expected, long desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __local *object, long __local *expected, long desired,
+    memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __global *object, long __private *expected,
+    long desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __local *object, long __private *expected,
+    long desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __global *object, ulong __global *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __local *object, ulong __global *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __global *object, ulong __local *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __local *object, ulong __local *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __global *object, ulong __private *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __local *object, ulong __private *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __global *object, ulong __global *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __local *object, ulong __global *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __global *object, ulong __local *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __local *object, ulong __local *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __global *object, ulong __private *expected,
+    ulong desired, memory_order success, memory_order failure);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __local *object, ulong __private *expected,
+    ulong desired, memory_order success, memory_order failure);
+#endif // __opencl_c_atomic_scope_device
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __global *object, long __global *expected,
+    long desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __local *object, long __global *expected, long desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __global *object, long __local *expected, long desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __local *object, long __local *expected, long desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __global *object, long __private *expected,
+    long desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_long __local *object, long __private *expected,
+    long desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __global *object, long __global *expected,
+    long desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __local *object, long __global *expected, long desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __global *object, long __local *expected, long desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __local *object, long __local *expected, long desired,
+    memory_order success, memory_order failure, memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __global *object, long __private *expected,
+    long desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_long __local *object, long __private *expected,
+    long desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __global *object, ulong __global *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __local *object, ulong __global *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __global *object, ulong __local *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __local *object, ulong __local *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __global *object, ulong __private *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_strong_explicit(
+    volatile atomic_ulong __local *object, ulong __private *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __global *object, ulong __global *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __local *object, ulong __global *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __global *object, ulong __local *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __local *object, ulong __local *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __global *object, ulong __private *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+bool __ovld atomic_compare_exchange_weak_explicit(
+    volatile atomic_ulong __local *object, ulong __private *expected,
+    ulong desired, memory_order success, memory_order failure,
+    memory_scope scope);
+#endif // defined(cl_khr_int64_base_atomics) &&
+       // defined(cl_khr_int64_extended_atomics)
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
 
 // atomic_flag_test_and_set() and atomic_flag_clear()
-
+#if defined(__opencl_c_atomic_scope_device) &&                                 \
+    defined(__opencl_c_atomic_order_seq_cst)
+#ifdef __opencl_c_generic_address_space
 bool __ovld atomic_flag_test_and_set(volatile atomic_flag *object);
-bool __ovld atomic_flag_test_and_set_explicit(volatile atomic_flag *object, memory_order order);
-bool __ovld atomic_flag_test_and_set_explicit(volatile atomic_flag *object, memory_order order, memory_scope scope);
 void __ovld atomic_flag_clear(volatile atomic_flag *object);
-void __ovld atomic_flag_clear_explicit(volatile atomic_flag *object, memory_order order);
-void __ovld atomic_flag_clear_explicit(volatile atomic_flag *object, memory_order order, memory_scope scope);
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+bool __ovld atomic_flag_test_and_set(volatile atomic_flag __global *object);
+bool __ovld atomic_flag_test_and_set(volatile atomic_flag __local *object);
+void __ovld atomic_flag_clear(volatile atomic_flag __global *object);
+void __ovld atomic_flag_clear(volatile atomic_flag __local *object);
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#endif // defined(__opencl_c_atomic_scope_device) &&
+       // defined(__opencl_c_atomic_order_seq_cst)
+
+#ifdef __opencl_c_generic_address_space
+#ifdef __opencl_c_atomic_scope_device
+bool __ovld atomic_flag_test_and_set_explicit(volatile atomic_flag *object,
+                                              memory_order order);
+void __ovld atomic_flag_clear_explicit(volatile atomic_flag *object,
+                                       memory_order order);
+#endif // __opencl_c_atomic_scope_device
+bool __ovld atomic_flag_test_and_set_explicit(volatile atomic_flag *object,
+                                              memory_order order,
+                                              memory_scope scope);
+void __ovld atomic_flag_clear_explicit(volatile atomic_flag *object,
+                                       memory_order order, memory_scope scope);
+#endif // __opencl_c_generic_address_space
+
+#if (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
+#ifdef __opencl_c_atomic_scope_device
+bool __ovld atomic_flag_test_and_set_explicit(
+    volatile atomic_flag __global *object, memory_order order);
+bool __ovld atomic_flag_test_and_set_explicit(
+    volatile atomic_flag __local *object, memory_order order);
+void __ovld atomic_flag_clear_explicit(volatile atomic_flag __global *object,
+                                       memory_order order);
+void __ovld atomic_flag_clear_explicit(volatile atomic_flag __local *object,
+                                       memory_order order);
+#endif // __opencl_c_atomic_scope_device
+bool __ovld
+atomic_flag_test_and_set_explicit(volatile atomic_flag __global *object,
+                                  memory_order order, memory_scope scope);
+bool __ovld
+atomic_flag_test_and_set_explicit(volatile atomic_flag __local *object,
+                                  memory_order order, memory_scope scope);
+void __ovld atomic_flag_clear_explicit(volatile atomic_flag __global *object,
+                                       memory_order order, memory_scope scope);
+void __ovld atomic_flag_clear_explicit(volatile atomic_flag __local *object,
+                                       memory_order order, memory_scope scope);
+#endif // (__OPENCL_C_VERSION__ >= CL_VERSION_3_0)
 
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
@@ -13918,7 +15943,7 @@ float16 __ovld __cnfn shuffle(float4 x, uint16 mask);
 float16 __ovld __cnfn shuffle(float8 x, uint16 mask);
 float16 __ovld __cnfn shuffle(float16 x, uint16 mask);
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double2 __ovld __cnfn shuffle(double2 x, ulong2 mask);
 double2 __ovld __cnfn shuffle(double4 x, ulong2 mask);
 double2 __ovld __cnfn shuffle(double8 x, ulong2 mask);
@@ -13938,7 +15963,7 @@ double16 __ovld __cnfn shuffle(double2 x, ulong16 mask);
 double16 __ovld __cnfn shuffle(double4 x, ulong16 mask);
 double16 __ovld __cnfn shuffle(double8 x, ulong16 mask);
 double16 __ovld __cnfn shuffle(double16 x, ulong16 mask);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #ifdef cl_khr_fp16
 half2 __ovld __cnfn shuffle(half2 x, ushort2 mask);
@@ -14142,7 +16167,7 @@ float16 __ovld __cnfn shuffle2(float4 x, float4 y, uint16 mask);
 float16 __ovld __cnfn shuffle2(float8 x, float8 y, uint16 mask);
 float16 __ovld __cnfn shuffle2(float16 x, float16 y, uint16 mask);
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double2 __ovld __cnfn shuffle2(double2 x, double2 y, ulong2 mask);
 double2 __ovld __cnfn shuffle2(double4 x, double4 y, ulong2 mask);
 double2 __ovld __cnfn shuffle2(double8 x, double8 y, ulong2 mask);
@@ -14162,7 +16187,7 @@ double16 __ovld __cnfn shuffle2(double2 x, double2 y, ulong16 mask);
 double16 __ovld __cnfn shuffle2(double4 x, double4 y, ulong16 mask);
 double16 __ovld __cnfn shuffle2(double8 x, double8 y, ulong16 mask);
 double16 __ovld __cnfn shuffle2(double16 x, double16 y, ulong16 mask);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #ifdef cl_khr_fp16
 half2 __ovld __cnfn shuffle2(half2 x, half2 y, ushort2 mask);
@@ -14198,6 +16223,7 @@ int printf(__constant const char* st, ...) __attribute__((format(printf, 1, 2)))
 #pragma OPENCL EXTENSION cl_khr_gl_msaa_sharing : enable
 #endif //cl_khr_gl_msaa_sharing
 
+#ifdef __opencl_c_images
 /**
  * Use the coordinate (coord.xy) to do an element lookup in
  * the 2D image object specified by image.
@@ -14476,6 +16502,7 @@ half4 __purefn __ovld read_imageh(read_only image1d_buffer_t image, int coord);
 
 // Image read functions for read_write images
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 float4 __purefn __ovld read_imagef(read_write image1d_t image, int coord);
 int4 __purefn __ovld read_imagei(read_write image1d_t image, int coord);
 uint4 __purefn __ovld read_imageui(read_write image1d_t image, int coord);
@@ -14519,6 +16546,7 @@ float __purefn __ovld read_imagef(read_write image2d_array_msaa_depth_t image, i
 #endif //cl_khr_gl_msaa_sharing
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 #ifdef cl_khr_mipmap_image
 float4 __purefn __ovld read_imagef(read_write image1d_t image, sampler_t sampler, float coord, float lod);
 int4 __purefn __ovld read_imagei(read_write image1d_t image, sampler_t sampler, float coord, float lod);
@@ -14569,6 +16597,7 @@ int4 __purefn __ovld read_imagei(read_write image3d_t image, sampler_t sampler,
 uint4 __purefn __ovld read_imageui(read_write image3d_t image, sampler_t sampler, float4 coord, float4 gradientX, float4 gradientY);
 
 #endif //cl_khr_mipmap_image
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 // Image read functions returning half4 type
@@ -14580,6 +16609,7 @@ half4 __purefn __ovld read_imageh(read_write image1d_array_t image, int2 coord);
 half4 __purefn __ovld read_imageh(read_write image2d_array_t image, int4 coord);
 half4 __purefn __ovld read_imageh(read_write image1d_buffer_t image, int coord);
 #endif //cl_khr_fp16
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
@@ -14669,7 +16699,7 @@ void __ovld write_imagef(write_only image1d_array_t image_array, int2 coord, flo
 void __ovld write_imagei(write_only image1d_array_t image_array, int2 coord, int4 color);
 void __ovld write_imageui(write_only image1d_array_t image_array, int2 coord, uint4 color);
 
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 void __ovld write_imagef(write_only image3d_t image, int4 coord, float4 color);
 void __ovld write_imagei(write_only image3d_t image, int4 coord, int4 color);
 void __ovld write_imageui(write_only image3d_t image, int4 coord, uint4 color);
@@ -14702,7 +16732,7 @@ void __ovld write_imageui(write_only image2d_array_t image_array, int4 coord, in
 void __ovld write_imagef(write_only image2d_depth_t image, int2 coord, int lod, float depth);
 void __ovld write_imagef(write_only image2d_array_depth_t image, int4 coord, int lod, float depth);
 
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 void __ovld write_imagef(write_only image3d_t image, int4 coord, int lod, float4 color);
 void __ovld write_imagei(write_only image3d_t image, int4 coord, int lod, int4 color);
 void __ovld write_imageui(write_only image3d_t image, int4 coord, int lod, uint4 color);
@@ -14715,7 +16745,7 @@ void __ovld write_imageui(write_only image3d_t image, int4 coord, int lod, uint4
 #ifdef cl_khr_fp16
 void __ovld write_imageh(write_only image1d_t image, int coord, half4 color);
 void __ovld write_imageh(write_only image2d_t image, int2 coord, half4 color);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 void __ovld write_imageh(write_only image3d_t image, int4 coord, half4 color);
 #endif
 void __ovld write_imageh(write_only image1d_array_t image, int2 coord, half4 color);
@@ -14725,6 +16755,7 @@ void __ovld write_imageh(write_only image1d_buffer_t image, int coord, half4 col
 
 // Image write functions for read_write images
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 void __ovld write_imagef(read_write image2d_t image, int2 coord, float4 color);
 void __ovld write_imagei(read_write image2d_t image, int2 coord, int4 color);
 void __ovld write_imageui(read_write image2d_t image, int2 coord, uint4 color);
@@ -14745,7 +16776,7 @@ void __ovld write_imagef(read_write image1d_array_t image_array, int2 coord, flo
 void __ovld write_imagei(read_write image1d_array_t image_array, int2 coord, int4 color);
 void __ovld write_imageui(read_write image1d_array_t image_array, int2 coord, uint4 color);
 
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 void __ovld write_imagef(read_write image3d_t image, int4 coord, float4 color);
 void __ovld write_imagei(read_write image3d_t image, int4 coord, int4 color);
 void __ovld write_imageui(read_write image3d_t image, int4 coord, uint4 color);
@@ -14777,7 +16808,7 @@ void __ovld write_imageui(read_write image2d_array_t image_array, int4 coord, in
 void __ovld write_imagef(read_write image2d_depth_t image, int2 coord, int lod, float color);
 void __ovld write_imagef(read_write image2d_array_depth_t image, int4 coord, int lod, float color);
 
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 void __ovld write_imagef(read_write image3d_t image, int4 coord, int lod, float4 color);
 void __ovld write_imagei(read_write image3d_t image, int4 coord, int lod, int4 color);
 void __ovld write_imageui(read_write image3d_t image, int4 coord, int lod, uint4 color);
@@ -14790,13 +16821,14 @@ void __ovld write_imageui(read_write image3d_t image, int4 coord, int lod, uint4
 #ifdef cl_khr_fp16
 void __ovld write_imageh(read_write image1d_t image, int coord, half4 color);
 void __ovld write_imageh(read_write image2d_t image, int2 coord, half4 color);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 void __ovld write_imageh(read_write image3d_t image, int4 coord, half4 color);
 #endif
 void __ovld write_imageh(read_write image1d_array_t image, int2 coord, half4 color);
 void __ovld write_imageh(read_write image2d_array_t image, int4 coord, half4 color);
 void __ovld write_imageh(read_write image1d_buffer_t image, int coord, half4 color);
 #endif //cl_khr_fp16
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 // Note: In OpenCL v1.0/1.1/1.2, image argument of image query builtin functions does not have
@@ -14810,7 +16842,7 @@ void __ovld write_imageh(read_write image1d_buffer_t image, int coord, half4 col
 int __ovld __cnfn get_image_width(read_only image1d_t image);
 int __ovld __cnfn get_image_width(read_only image1d_buffer_t image);
 int __ovld __cnfn get_image_width(read_only image2d_t image);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 int __ovld __cnfn get_image_width(read_only image3d_t image);
 #endif
 int __ovld __cnfn get_image_width(read_only image1d_array_t image);
@@ -14829,7 +16861,7 @@ int __ovld __cnfn get_image_width(read_only image2d_array_msaa_depth_t image);
 int __ovld __cnfn get_image_width(write_only image1d_t image);
 int __ovld __cnfn get_image_width(write_only image1d_buffer_t image);
 int __ovld __cnfn get_image_width(write_only image2d_t image);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 int __ovld __cnfn get_image_width(write_only image3d_t image);
 #endif
 int __ovld __cnfn get_image_width(write_only image1d_array_t image);
@@ -14846,6 +16878,7 @@ int __ovld __cnfn get_image_width(write_only image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 int __ovld __cnfn get_image_width(read_write image1d_t image);
 int __ovld __cnfn get_image_width(read_write image1d_buffer_t image);
 int __ovld __cnfn get_image_width(read_write image2d_t image);
@@ -14862,6 +16895,7 @@ int __ovld __cnfn get_image_width(read_write image2d_msaa_depth_t image);
 int __ovld __cnfn get_image_width(read_write image2d_array_msaa_t image);
 int __ovld __cnfn get_image_width(read_write image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
@@ -14882,7 +16916,7 @@ int __ovld __cnfn get_image_height(read_only image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
 
 int __ovld __cnfn get_image_height(write_only image2d_t image);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 int __ovld __cnfn get_image_height(write_only image3d_t image);
 #endif
 int __ovld __cnfn get_image_height(write_only image2d_array_t image);
@@ -14898,6 +16932,7 @@ int __ovld __cnfn get_image_height(write_only image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 int __ovld __cnfn get_image_height(read_write image2d_t image);
 int __ovld __cnfn get_image_height(read_write image3d_t image);
 int __ovld __cnfn get_image_height(read_write image2d_array_t image);
@@ -14911,6 +16946,7 @@ int __ovld __cnfn get_image_height(read_write image2d_msaa_depth_t image);
 int __ovld __cnfn get_image_height(read_write image2d_array_msaa_t image);
 int __ovld __cnfn get_image_height(read_write image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
@@ -14918,12 +16954,14 @@ int __ovld __cnfn get_image_height(read_write image2d_array_msaa_depth_t image);
  */
 int __ovld __cnfn get_image_depth(read_only image3d_t image);
 
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 int __ovld __cnfn get_image_depth(write_only image3d_t image);
 #endif
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 int __ovld __cnfn get_image_depth(read_write image3d_t image);
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 // OpenCL Extension v2.0 s9.18 - Mipmaps
@@ -14939,13 +16977,15 @@ int __ovld get_image_num_mip_levels(read_only image3d_t image);
 
 int __ovld get_image_num_mip_levels(write_only image1d_t image);
 int __ovld get_image_num_mip_levels(write_only image2d_t image);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 int __ovld get_image_num_mip_levels(write_only image3d_t image);
 #endif
 
+#ifdef __opencl_c_read_write_images
 int __ovld get_image_num_mip_levels(read_write image1d_t image);
 int __ovld get_image_num_mip_levels(read_write image2d_t image);
 int __ovld get_image_num_mip_levels(read_write image3d_t image);
+#endif //__opencl_c_read_write_images
 
 int __ovld get_image_num_mip_levels(read_only image1d_array_t image);
 int __ovld get_image_num_mip_levels(read_only image2d_array_t image);
@@ -14957,10 +16997,12 @@ int __ovld get_image_num_mip_levels(write_only image2d_array_t image);
 int __ovld get_image_num_mip_levels(write_only image2d_array_depth_t image);
 int __ovld get_image_num_mip_levels(write_only image2d_depth_t image);
 
+#ifdef __opencl_c_read_write_images
 int __ovld get_image_num_mip_levels(read_write image1d_array_t image);
 int __ovld get_image_num_mip_levels(read_write image2d_array_t image);
 int __ovld get_image_num_mip_levels(read_write image2d_array_depth_t image);
 int __ovld get_image_num_mip_levels(read_write image2d_depth_t image);
+#endif //__opencl_c_read_write_images
 
 #endif //cl_khr_mipmap_image
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
@@ -15004,7 +17046,7 @@ int __ovld __cnfn get_image_channel_data_type(read_only image2d_array_msaa_depth
 int __ovld __cnfn get_image_channel_data_type(write_only image1d_t image);
 int __ovld __cnfn get_image_channel_data_type(write_only image1d_buffer_t image);
 int __ovld __cnfn get_image_channel_data_type(write_only image2d_t image);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 int __ovld __cnfn get_image_channel_data_type(write_only image3d_t image);
 #endif
 int __ovld __cnfn get_image_channel_data_type(write_only image1d_array_t image);
@@ -15021,6 +17063,7 @@ int __ovld __cnfn get_image_channel_data_type(write_only image2d_array_msaa_dept
 #endif //cl_khr_gl_msaa_sharing
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 int __ovld __cnfn get_image_channel_data_type(read_write image1d_t image);
 int __ovld __cnfn get_image_channel_data_type(read_write image1d_buffer_t image);
 int __ovld __cnfn get_image_channel_data_type(read_write image2d_t image);
@@ -15037,6 +17080,7 @@ int __ovld __cnfn get_image_channel_data_type(read_write image2d_msaa_depth_t im
 int __ovld __cnfn get_image_channel_data_type(read_write image2d_array_msaa_t image);
 int __ovld __cnfn get_image_channel_data_type(read_write image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
@@ -15076,7 +17120,7 @@ int __ovld __cnfn get_image_channel_order(read_only image2d_array_msaa_depth_t i
 int __ovld __cnfn get_image_channel_order(write_only image1d_t image);
 int __ovld __cnfn get_image_channel_order(write_only image1d_buffer_t image);
 int __ovld __cnfn get_image_channel_order(write_only image2d_t image);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 int __ovld __cnfn get_image_channel_order(write_only image3d_t image);
 #endif
 int __ovld __cnfn get_image_channel_order(write_only image1d_array_t image);
@@ -15093,6 +17137,7 @@ int __ovld __cnfn get_image_channel_order(write_only image2d_array_msaa_depth_t
 #endif //cl_khr_gl_msaa_sharing
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 int __ovld __cnfn get_image_channel_order(read_write image1d_t image);
 int __ovld __cnfn get_image_channel_order(read_write image1d_buffer_t image);
 int __ovld __cnfn get_image_channel_order(read_write image2d_t image);
@@ -15109,6 +17154,7 @@ int __ovld __cnfn get_image_channel_order(read_write image2d_msaa_depth_t image)
 int __ovld __cnfn get_image_channel_order(read_write image2d_array_msaa_t image);
 int __ovld __cnfn get_image_channel_order(read_write image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
@@ -15143,6 +17189,7 @@ int2 __ovld __cnfn get_image_dim(write_only image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 int2 __ovld __cnfn get_image_dim(read_write image2d_t image);
 int2 __ovld __cnfn get_image_dim(read_write image2d_array_t image);
 #ifdef cl_khr_depth_images
@@ -15155,6 +17202,7 @@ int2 __ovld __cnfn get_image_dim(read_write image2d_msaa_depth_t image);
 int2 __ovld __cnfn get_image_dim(read_write image2d_array_msaa_t image);
 int2 __ovld __cnfn get_image_dim(read_write image2d_array_msaa_depth_t image);
 #endif //cl_khr_gl_msaa_sharing
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
@@ -15164,11 +17212,13 @@ int2 __ovld __cnfn get_image_dim(read_write image2d_array_msaa_depth_t image);
  * component and the w component is 0.
  */
 int4 __ovld __cnfn get_image_dim(read_only image3d_t image);
-#ifdef cl_khr_3d_image_writes
+#if defined(cl_khr_3d_image_writes) || defined(__opencl_c_3d_image_writes)
 int4 __ovld __cnfn get_image_dim(write_only image3d_t image);
 #endif
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 int4 __ovld __cnfn get_image_dim(read_write image3d_t image);
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
@@ -15196,6 +17246,7 @@ size_t __ovld __cnfn get_image_array_size(write_only image2d_array_msaa_depth_t
 #endif //cl_khr_gl_msaa_sharing
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 size_t __ovld __cnfn get_image_array_size(read_write image1d_array_t image_array);
 size_t __ovld __cnfn get_image_array_size(read_write image2d_array_t image_array);
 #ifdef cl_khr_depth_images
@@ -15205,6 +17256,7 @@ size_t __ovld __cnfn get_image_array_size(read_write image2d_array_depth_t image
 size_t __ovld __cnfn get_image_array_size(read_write image2d_array_msaa_t image_array);
 size_t __ovld __cnfn get_image_array_size(read_write image2d_array_msaa_depth_t image_array);
 #endif //cl_khr_gl_msaa_sharing
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 /**
@@ -15222,16 +17274,21 @@ int __ovld get_image_num_samples(write_only image2d_array_msaa_t image);
 int __ovld get_image_num_samples(write_only image2d_array_msaa_depth_t image);
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_read_write_images
 int __ovld get_image_num_samples(read_write image2d_msaa_t image);
 int __ovld get_image_num_samples(read_write image2d_msaa_depth_t image);
 int __ovld get_image_num_samples(read_write image2d_array_msaa_t image);
 int __ovld get_image_num_samples(read_write image2d_array_msaa_depth_t image);
+#endif //__opencl_c_read_write_images
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 #endif
 
+#endif //__opencl_c_images
+
 // OpenCL v2.0 s6.13.15 - Work-group Functions
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_work_group_collective_functions
 int __ovld __conv work_group_all(int predicate);
 int __ovld __conv work_group_any(int predicate);
 
@@ -15255,11 +17312,11 @@ ulong __ovld __conv work_group_broadcast(ulong a, size_t x, size_t y, size_t z);
 float __ovld __conv work_group_broadcast(float a, size_t local_id);
 float __ovld __conv work_group_broadcast(float a, size_t x, size_t y);
 float __ovld __conv work_group_broadcast(float a, size_t x, size_t y, size_t z);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __conv work_group_broadcast(double a, size_t local_id);
 double __ovld __conv work_group_broadcast(double a, size_t x, size_t y);
 double __ovld __conv work_group_broadcast(double a, size_t x, size_t y, size_t z);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #ifdef cl_khr_fp16
 half __ovld __conv work_group_reduce_add(half x);
@@ -15317,7 +17374,7 @@ float __ovld __conv work_group_scan_exclusive_max(float x);
 float __ovld __conv work_group_scan_inclusive_add(float x);
 float __ovld __conv work_group_scan_inclusive_min(float x);
 float __ovld __conv work_group_scan_inclusive_max(float x);
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double __ovld __conv work_group_reduce_add(double x);
 double __ovld __conv work_group_reduce_min(double x);
 double __ovld __conv work_group_reduce_max(double x);
@@ -15327,19 +17384,18 @@ double __ovld __conv work_group_scan_exclusive_max(double x);
 double __ovld __conv work_group_scan_inclusive_add(double x);
 double __ovld __conv work_group_scan_inclusive_min(double x);
 double __ovld __conv work_group_scan_inclusive_max(double x);
-#endif //cl_khr_fp64
-
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
+#endif //__opencl_c_work_group_collective_functions
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 // OpenCL v2.0 s6.13.16 - Pipe Functions
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_pipes
 bool __ovld is_valid_reserve_id(reserve_id_t reserve_id);
-#endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
-
+#endif //__opencl_c_pipes
 
 // OpenCL v2.0 s6.13.17 - Enqueue Kernels
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
-
+#ifdef __opencl_c_device_enqueue
 ndrange_t __ovld ndrange_1D(size_t);
 ndrange_t __ovld ndrange_1D(size_t, size_t);
 ndrange_t __ovld ndrange_1D(size_t, size_t, size_t);
@@ -15367,11 +17423,13 @@ bool __ovld is_valid_event (clk_event_t event);
 void __ovld capture_event_profiling_info(clk_event_t, clk_profiling_info, __global void* value);
 
 queue_t __ovld get_default_queue(void);
+#endif //__opencl_c_device_enqueue
 #endif //defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 // OpenCL Extension v2.0 s9.17 - Sub-groups
 
-#if defined(cl_intel_subgroups) || defined(cl_khr_subgroups)
+#if defined(cl_intel_subgroups) || defined(cl_khr_subgroups) ||                \
+    defined(__opencl_c_subgroups)
 // Shared Sub Group Functions
 uint    __ovld get_sub_group_size(void);
 uint    __ovld get_max_sub_group_size(void);
@@ -15457,7 +17515,7 @@ half    __ovld __conv sub_group_scan_inclusive_min(half x);
 half    __ovld __conv sub_group_scan_inclusive_max(half x);
 #endif //cl_khr_fp16
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double  __ovld __conv sub_group_broadcast(double x, uint sub_group_local_id);
 double  __ovld __conv sub_group_reduce_add(double x);
 double  __ovld __conv sub_group_reduce_min(double x);
@@ -15468,7 +17526,7 @@ double  __ovld __conv sub_group_scan_exclusive_max(double x);
 double  __ovld __conv sub_group_scan_inclusive_add(double x);
 double  __ovld __conv sub_group_scan_inclusive_min(double x);
 double  __ovld __conv sub_group_scan_inclusive_max(double x);
-#endif //cl_khr_fp64
+#endif // defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 
 #endif //cl_khr_subgroups cl_intel_subgroups
 
@@ -15570,16 +17628,22 @@ uint16  __ovld __conv intel_sub_group_shuffle_xor( uint16 x, uint c );
 long    __ovld __conv intel_sub_group_shuffle_xor( long x, uint c );
 ulong   __ovld __conv intel_sub_group_shuffle_xor( ulong x, uint c );
 
+#ifdef __opencl_c_images
 uint    __ovld __conv intel_sub_group_block_read( read_only image2d_t image, int2 coord );
 uint2   __ovld __conv intel_sub_group_block_read2( read_only image2d_t image, int2 coord );
 uint4   __ovld __conv intel_sub_group_block_read4( read_only image2d_t image, int2 coord );
 uint8   __ovld __conv intel_sub_group_block_read8( read_only image2d_t image, int2 coord );
+#endif //__opencl_c_images
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_images
+#ifdef __opencl_c_read_write_images
 uint    __ovld __conv intel_sub_group_block_read(read_write image2d_t image, int2 coord);
 uint2   __ovld __conv intel_sub_group_block_read2(read_write image2d_t image, int2 coord);
 uint4   __ovld __conv intel_sub_group_block_read4(read_write image2d_t image, int2 coord);
 uint8   __ovld __conv intel_sub_group_block_read8(read_write image2d_t image, int2 coord);
+#endif //__opencl_c_read_write_images
+#endif //__opencl_c_images
 #endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 uint    __ovld __conv intel_sub_group_block_read( const __global uint* p );
@@ -15587,16 +17651,22 @@ uint2   __ovld __conv intel_sub_group_block_read2( const __global uint* p );
 uint4   __ovld __conv intel_sub_group_block_read4( const __global uint* p );
 uint8   __ovld __conv intel_sub_group_block_read8( const __global uint* p );
 
+#ifdef __opencl_c_images
 void    __ovld __conv intel_sub_group_block_write(write_only image2d_t image, int2 coord, uint data);
 void    __ovld __conv intel_sub_group_block_write2(write_only image2d_t image, int2 coord, uint2 data);
 void    __ovld __conv intel_sub_group_block_write4(write_only image2d_t image, int2 coord, uint4 data);
 void    __ovld __conv intel_sub_group_block_write8(write_only image2d_t image, int2 coord, uint8 data);
+#endif //__opencl_c_images
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_images
+#ifdef __opencl_c_read_write_images
 void    __ovld __conv intel_sub_group_block_write(read_write image2d_t image, int2 coord, uint data);
 void    __ovld __conv intel_sub_group_block_write2(read_write image2d_t image, int2 coord, uint2 data);
 void    __ovld __conv intel_sub_group_block_write4(read_write image2d_t image, int2 coord, uint4 data);
 void    __ovld __conv intel_sub_group_block_write8(read_write image2d_t image, int2 coord, uint8 data);
+#endif // __opencl_c_read_write_images
+#endif // __opencl_c_images
 #endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 void    __ovld __conv intel_sub_group_block_write( __global uint* p, uint data );
@@ -15611,7 +17681,7 @@ half    __ovld __conv intel_sub_group_shuffle_up( half prev, half cur, uint c );
 half    __ovld __conv intel_sub_group_shuffle_xor( half x, uint c );
 #endif
 
-#if defined(cl_khr_fp64)
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 double  __ovld __conv intel_sub_group_shuffle( double x, uint c );
 double  __ovld __conv intel_sub_group_shuffle_down( double prev, double cur, uint c );
 double  __ovld __conv intel_sub_group_shuffle_up( double prev, double cur, uint c );
@@ -15710,16 +17780,22 @@ ushort      __ovld __conv intel_sub_group_scan_inclusive_min( ushort  x );
 short       __ovld __conv intel_sub_group_scan_inclusive_max( short   x );
 ushort      __ovld __conv intel_sub_group_scan_inclusive_max( ushort  x );
 
+#ifdef __opencl_c_images
 uint       __ovld __conv intel_sub_group_block_read_ui( read_only image2d_t image, int2 byte_coord );
 uint2      __ovld __conv intel_sub_group_block_read_ui2( read_only image2d_t image, int2 byte_coord );
 uint4      __ovld __conv intel_sub_group_block_read_ui4( read_only image2d_t image, int2 byte_coord );
 uint8      __ovld __conv intel_sub_group_block_read_ui8( read_only image2d_t image, int2 byte_coord );
+#endif //__opencl_c_images
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_images
+#ifdef __opencl_c_read_write_images
 uint       __ovld __conv intel_sub_group_block_read_ui( read_write image2d_t image, int2 byte_coord );
 uint2      __ovld __conv intel_sub_group_block_read_ui2( read_write image2d_t image, int2 byte_coord );
 uint4      __ovld __conv intel_sub_group_block_read_ui4( read_write image2d_t image, int2 byte_coord );
 uint8      __ovld __conv intel_sub_group_block_read_ui8( read_write image2d_t image, int2 byte_coord );
+#endif //__opencl_c_read_write_images
+#endif //__opencl_c_images
 #endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 uint       __ovld __conv intel_sub_group_block_read_ui( const __global uint* p );
@@ -15727,16 +17803,22 @@ uint2      __ovld __conv intel_sub_group_block_read_ui2( const __global uint* p
 uint4      __ovld __conv intel_sub_group_block_read_ui4( const __global uint* p );
 uint8      __ovld __conv intel_sub_group_block_read_ui8( const __global uint* p );
 
+#ifdef __opencl_c_images
 void       __ovld __conv intel_sub_group_block_write_ui( read_only image2d_t image, int2 byte_coord, uint data );
 void       __ovld __conv intel_sub_group_block_write_ui2( read_only image2d_t image, int2 byte_coord, uint2 data );
 void       __ovld __conv intel_sub_group_block_write_ui4( read_only image2d_t image, int2 byte_coord, uint4 data );
 void       __ovld __conv intel_sub_group_block_write_ui8( read_only image2d_t image, int2 byte_coord, uint8 data );
+#endif //__opencl_c_images
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_images
+#ifdef __opencl_c_read_write_images
 void       __ovld __conv intel_sub_group_block_write_ui( read_write image2d_t image, int2 byte_coord, uint data );
 void       __ovld __conv intel_sub_group_block_write_ui2( read_write image2d_t image, int2 byte_coord, uint2 data );
 void       __ovld __conv intel_sub_group_block_write_ui4( read_write image2d_t image, int2 byte_coord, uint4 data );
 void       __ovld __conv intel_sub_group_block_write_ui8( read_write image2d_t image, int2 byte_coord, uint8 data );
+#endif //__opencl_c_read_write_images
+#endif //__opencl_c_images
 #endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 void       __ovld __conv intel_sub_group_block_write_ui( __global uint* p, uint data );
@@ -15744,16 +17826,22 @@ void       __ovld __conv intel_sub_group_block_write_ui2( __global uint* p, uint
 void       __ovld __conv intel_sub_group_block_write_ui4( __global uint* p, uint4 data );
 void       __ovld __conv intel_sub_group_block_write_ui8( __global uint* p, uint8 data );
 
+#ifdef __opencl_c_images
 ushort      __ovld __conv intel_sub_group_block_read_us( read_only image2d_t image, int2 coord );
 ushort2     __ovld __conv intel_sub_group_block_read_us2( read_only image2d_t image, int2 coord );
 ushort4     __ovld __conv intel_sub_group_block_read_us4( read_only image2d_t image, int2 coord );
 ushort8     __ovld __conv intel_sub_group_block_read_us8( read_only image2d_t image, int2 coord );
+#endif //__opencl_c_images
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_images
+#ifdef __opencl_c_read_write_images
 ushort      __ovld __conv intel_sub_group_block_read_us(read_write image2d_t image, int2 coord);
 ushort2     __ovld __conv intel_sub_group_block_read_us2(read_write image2d_t image, int2 coord);
 ushort4     __ovld __conv intel_sub_group_block_read_us4(read_write image2d_t image, int2 coord);
 ushort8     __ovld __conv intel_sub_group_block_read_us8(read_write image2d_t image, int2 coord);
+#endif //__opencl_c_read_write_images
+#endif //__opencl_c_images
 #endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 ushort      __ovld __conv intel_sub_group_block_read_us(  const __global ushort* p );
@@ -15761,16 +17849,22 @@ ushort2     __ovld __conv intel_sub_group_block_read_us2( const __global ushort*
 ushort4     __ovld __conv intel_sub_group_block_read_us4( const __global ushort* p );
 ushort8     __ovld __conv intel_sub_group_block_read_us8( const __global ushort* p );
 
+#ifdef __opencl_c_images
 void        __ovld __conv intel_sub_group_block_write_us(write_only image2d_t image, int2 coord, ushort  data);
 void        __ovld __conv intel_sub_group_block_write_us2(write_only image2d_t image, int2 coord, ushort2 data);
 void        __ovld __conv intel_sub_group_block_write_us4(write_only image2d_t image, int2 coord, ushort4 data);
 void        __ovld __conv intel_sub_group_block_write_us8(write_only image2d_t image, int2 coord, ushort8 data);
+#endif //__opencl_c_images
 
 #if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#ifdef __opencl_c_images
+#ifdef __opencl_c_read_write_images
 void        __ovld __conv intel_sub_group_block_write_us(read_write image2d_t image, int2 coord, ushort  data);
 void        __ovld __conv intel_sub_group_block_write_us2(read_write image2d_t image, int2 coord, ushort2 data);
 void        __ovld __conv intel_sub_group_block_write_us4(read_write image2d_t image, int2 coord, ushort4 data);
 void        __ovld __conv intel_sub_group_block_write_us8(read_write image2d_t image, int2 coord, ushort8 data);
+#endif //__opencl_c_read_write_images
+#endif //__opencl_c_images
 #endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
 
 void        __ovld __conv intel_sub_group_block_write_us(  __global ushort* p, ushort  data );
@@ -15889,6 +17983,7 @@ short2 __ovld intel_sub_group_avc_ime_adjust_ref_offset(
     short2 ref_offset, ushort2 src_coord, ushort2 ref_window_size,
     ushort2 image_size);
 
+#ifdef __opencl_c_images
 intel_sub_group_avc_ime_result_t __ovld
 intel_sub_group_avc_ime_evaluate_with_single_reference(
     read_only image2d_t src_image, read_only image2d_t ref_image,
@@ -15929,6 +18024,7 @@ intel_sub_group_avc_ime_evaluate_with_dual_reference_streaminout(
     read_only image2d_t bwd_ref_image, sampler_t vme_media_sampler,
     intel_sub_group_avc_ime_payload_t payload,
     intel_sub_group_avc_ime_dual_reference_streamin_t streamin_components);
+#endif //__opencl_c_images
 
 intel_sub_group_avc_ime_single_reference_streamin_t __ovld
 intel_sub_group_avc_ime_get_single_reference_streamin(
@@ -15993,6 +18089,7 @@ intel_sub_group_avc_ref_payload_t __ovld
 intel_sub_group_avc_ref_set_bilinear_filter_enable(
     intel_sub_group_avc_ref_payload_t payload);
 
+#ifdef __opencl_c_images
 intel_sub_group_avc_ref_result_t __ovld
 intel_sub_group_avc_ref_evaluate_with_single_reference(
     read_only image2d_t src_image, read_only image2d_t ref_image,
@@ -16011,6 +18108,7 @@ intel_sub_group_avc_ref_evaluate_with_multi_reference(
     read_only image2d_t src_image, uint packed_reference_ids,
     uchar packed_reference_field_polarities, sampler_t vme_media_sampler,
     intel_sub_group_avc_ref_payload_t payload);
+#endif //__opencl_c_images
 
 // SIC built-in functions
 intel_sub_group_avc_sic_payload_t __ovld
@@ -16061,6 +18159,7 @@ intel_sub_group_avc_sic_set_block_based_raw_skip_sad(
     uchar block_based_skip_type,
     intel_sub_group_avc_sic_payload_t payload);
 
+#ifdef __opencl_c_images
 intel_sub_group_avc_sic_result_t __ovld
 intel_sub_group_avc_sic_evaluate_ipe(
     read_only image2d_t src_image, sampler_t vme_media_sampler,
@@ -16083,6 +18182,7 @@ intel_sub_group_avc_sic_evaluate_with_multi_reference(
     read_only image2d_t src_image, uint packed_reference_ids,
     uchar packed_reference_field_polarities, sampler_t vme_media_sampler,
     intel_sub_group_avc_sic_payload_t payload);
+#endif //__opencl_c_images
 
 uchar __ovld intel_sub_group_avc_sic_get_ipe_luma_shape(
     intel_sub_group_avc_sic_result_t result);
diff --git a/clang/lib/Parse/ParseDecl.cpp b/clang/lib/Parse/ParseDecl.cpp
index 6353e14bc41a..a8c216de6e04 100644
--- a/clang/lib/Parse/ParseDecl.cpp
+++ b/clang/lib/Parse/ParseDecl.cpp
@@ -3904,8 +3904,7 @@ void Parser::ParseDeclarationSpecifiers(DeclSpec &DS,
       isInvalid = DS.SetTypeAltiVecBool(true, Loc, PrevSpec, DiagID, Policy);
       break;
     case tok::kw_pipe:
-      if (!getLangOpts().OpenCL || (getLangOpts().OpenCLVersion < 200 &&
-                                    !getLangOpts().OpenCLCPlusPlus)) {
+      if (!getLangOpts().OpenCLPipeKeyword) {
         // OpenCL 2.0 defined this keyword. OpenCL 1.2 and earlier should
         // support the "pipe" word as identifier.
         Tok.getIdentifierInfo()->revertTokenIDToIdentifier();
@@ -4027,8 +4026,7 @@ void Parser::ParseDeclarationSpecifiers(DeclSpec &DS,
     case tok::kw___generic:
       // generic address space is introduced only in OpenCL v2.0
       // see OpenCL C Spec v2.0 s6.5.5
-      if (Actions.getLangOpts().OpenCLVersion < 200 &&
-          !Actions.getLangOpts().OpenCLCPlusPlus) {
+      if (!Actions.getLangOpts().OpenCLGenericKeyword) {
         DiagID = diag::err_opencl_unknown_type_specifier;
         PrevSpec = Tok.getIdentifierInfo()->getNameStart();
         isInvalid = true;
@@ -5050,8 +5048,7 @@ bool Parser::isDeclarationSpecifier(bool DisambiguatingWithExpression) {
   default: return false;
 
   case tok::kw_pipe:
-    return (getLangOpts().OpenCL && getLangOpts().OpenCLVersion >= 200) ||
-           getLangOpts().OpenCLCPlusPlus;
+    return getLangOpts().OpenCLPipeKeyword;
 
   case tok::identifier:   // foo::bar
     // Unfortunate hack to support "Class.factoryMethod" notation.
@@ -5557,8 +5554,7 @@ static bool isPtrOperatorToken(tok::TokenKind Kind, const LangOptions &Lang,
   if (Kind == tok::star || Kind == tok::caret)
     return true;
 
-  if (Kind == tok::kw_pipe &&
-      ((Lang.OpenCL && Lang.OpenCLVersion >= 200) || Lang.OpenCLCPlusPlus))
+  if (Kind == tok::kw_pipe && Lang.OpenCLPipeKeyword)
     return true;
 
   if (!Lang.CPlusPlus)
diff --git a/clang/lib/Parse/ParsePragma.cpp b/clang/lib/Parse/ParsePragma.cpp
index df411e1928d6..9a40ce888695 100644
--- a/clang/lib/Parse/ParsePragma.cpp
+++ b/clang/lib/Parse/ParsePragma.cpp
@@ -697,12 +697,14 @@ void Parser::HandlePragmaOpenCLExtension() {
   if (Name == "all") {
     if (State == Disable) {
       Opt.disableAll();
-      Opt.enableSupportedCore(getLangOpts());
+      Opt.enableSupportedCore();
     } else {
       PP.Diag(NameLoc, diag::warn_pragma_expected_predicate) << 1;
     }
+  } else if (Opt.isFeature(Name)) {
+    PP.Diag(NameLoc, diag::warn_opencl_pragma_feature_ignore) << Ident;
   } else if (State == Begin) {
-    if (!Opt.isKnown(Name) || !Opt.isSupported(Name, getLangOpts())) {
+    if (!Opt.isKnown(Name) || !Opt.isSupported(Name)) {
       Opt.support(Name);
     }
     Actions.setCurrentOpenCLExtension(Name);
@@ -712,9 +714,9 @@ void Parser::HandlePragmaOpenCLExtension() {
     Actions.setCurrentOpenCLExtension("");
   } else if (!Opt.isKnown(Name))
     PP.Diag(NameLoc, diag::warn_pragma_unknown_extension) << Ident;
-  else if (Opt.isSupportedExtension(Name, getLangOpts()))
+  else if (Opt.isSupportedExtension(Name))
     Opt.enable(Name, State == Enable);
-  else if (Opt.isSupportedCore(Name, getLangOpts()))
+  else if (Opt.isSupportedCore(Name))
     PP.Diag(NameLoc, diag::warn_pragma_extension_is_core) << Ident;
   else
     PP.Diag(NameLoc, diag::warn_pragma_unsupported_extension) << Ident;
diff --git a/clang/lib/Sema/OpenCLBuiltins.td b/clang/lib/Sema/OpenCLBuiltins.td
index 9d6bb411eff8..d352d35f1e46 100644
--- a/clang/lib/Sema/OpenCLBuiltins.td
+++ b/clang/lib/Sema/OpenCLBuiltins.td
@@ -22,11 +22,13 @@
 class Version<int _Version> {
   int ID = _Version;
 }
+
 def CLAll : Version<  0>;
 def CL10  : Version<100>;
 def CL11  : Version<110>;
 def CL12  : Version<120>;
 def CL20  : Version<200>;
+def CL30  : Version<300>;
 
 // Address spaces
 // Pointer types need to be assigned an address space.
@@ -65,6 +67,14 @@ def FuncExtKhrGlMsaaSharing              : FunctionExtension<"cl_khr_gl_msaa_sha
 // Multiple extensions
 def FuncExtKhrMipmapAndWrite3d           : FunctionExtension<"cl_khr_mipmap_image cl_khr_3d_image_writes">;
 
+// Features
+def FuncExtGenericAddressSpace                : FunctionExtension<"__opencl_c_generic_address_space">;
+def FuncExtWorkGroupCollective                : FunctionExtension<"__opencl_c_work_group_collective_functions">;
+def FuncExtPipes                              : FunctionExtension<"__opencl_c_pipes">;
+def FuncExtDeviceSidEenqueue                  : FunctionExtension<"__opencl_c_device_enqueue">;
+def FuncNonExplicitAtomicFeatures             : FunctionExtension<"__opencl_c_atomic_order_seq_cst __opencl_c_atomic_scope_device">;
+def FuncNonExplicitAtomicFeaturesAndGenericAS : FunctionExtension<"__opencl_c_atomic_order_seq_cst __opencl_c_atomic_scope_device __opencl_c_generic_address_space">;
+
 // Qualified Type.  These map to ASTContext::QualType.
 class QualType<string _Name, bit _IsAbstract=0> {
   // Name of the field or function in a clang::ASTContext
@@ -230,13 +240,9 @@ class Builtin<string _Name, list<Type> _Signature, list<bit> _Attributes = Attr.
   bit IsConv = _Attributes[2];
   // OpenCL extensions to which the function belongs.
   FunctionExtension Extension = FuncExtNone;
-  // Version of OpenCL from which the function is available (e.g.: CL10).
-  // MinVersion is inclusive.
-  Version MinVersion = CL10;
-  // Version of OpenCL from which the function is not supported anymore.
-  // MaxVersion is exclusive.
+  // List of OpenCL version in which this function available.
   // CLAll makes the function available for all versions.
-  Version MaxVersion = CLAll;
+  list<Version> Versions = [CLAll];
 }
 
 //===----------------------------------------------------------------------===//
@@ -398,7 +404,7 @@ foreach name = ["get_global_size", "get_global_id", "get_local_size",
   def : Builtin<name, [Size, UInt], Attr.Const>;
 }
 
-let MinVersion = CL20 in {
+let Versions = [CL20, CL30] in {
   def : Builtin<"get_enqueued_local_size", [Size, UInt]>;
   foreach name = ["get_global_linear_id", "get_local_linear_id"] in {
     def : Builtin<name, [Size]>;
@@ -471,7 +477,7 @@ foreach name = ["fma", "mad"] in {
 }
 
 // --- Version dependent ---
-let MaxVersion = CL20 in {
+let Versions = [CL10, CL11, CL12, CL30] in {
   foreach AS = [GlobalAS, LocalAS, PrivateAS] in {
     foreach name = ["fract", "modf", "sincos"] in {
       def : Builtin<name, [FGenTypeN, FGenTypeN, PointerType<FGenTypeN, AS>]>;
@@ -488,7 +494,9 @@ let MaxVersion = CL20 in {
     }
   }
 }
-let MinVersion = CL20 in {
+
+let Versions = [CL20, CL30] in {
+ let Extension = FuncExtGenericAddressSpace in {
   foreach name = ["fract", "modf", "sincos"] in {
     def : Builtin<name, [FGenTypeN, FGenTypeN, PointerType<FGenTypeN, GenericAS>]>;
   }
@@ -501,6 +509,7 @@ let MinVersion = CL20 in {
       def : Builtin<name, [Type, Type, Type, PointerType<GenTypeIntVecAndScalar, GenericAS>]>;
     }
   }
+ }
 }
 
 // --- Table 9 ---
@@ -531,7 +540,7 @@ foreach name = ["abs"] in {
 foreach name = ["clz", "popcount"] in {
   def : Builtin<name, [AIGenTypeN, AIGenTypeN], Attr.Const>;
 }
-let MinVersion = CL20 in {
+let Versions = [CL20, CL30] in {
   foreach name = ["ctz"] in {
     def : Builtin<name, [AIGenTypeN, AIGenTypeN]>;
   }
@@ -705,7 +714,7 @@ foreach name = ["select"] in {
 // --- Table 15 ---
 // Variants for OpenCL versions below 2.0, using pointers to the global, local
 // and private address spaces.
-let MaxVersion = CL20 in {
+let Versions = [CL10, CL11, CL12, CL30] in {
   foreach AS = [GlobalAS, LocalAS, PrivateAS] in {
     foreach VSize = [2, 3, 4, 8, 16] in {
       foreach name = ["vload" # VSize] in {
@@ -748,7 +757,8 @@ let MaxVersion = CL20 in {
 }
 // Variants for OpenCL versions above 2.0, using pointers to the generic
 // address space.
-let MinVersion = CL20 in {
+let Versions = [CL20, CL30] in {
+ let Extension = FuncExtGenericAddressSpace in {
   foreach VSize = [2, 3, 4, 8, 16] in {
     foreach name = ["vload" # VSize] in {
       def : Builtin<name, [VectorType<Char, VSize>, Size, PointerType<ConstType<Char>, GenericAS>]>;
@@ -786,6 +796,7 @@ let MinVersion = CL20 in {
       }
     }
   }
+ }
 }
 // Variants using pointers to the constant address space.
 foreach VSize = [2, 3, 4, 8, 16] in {
@@ -812,7 +823,7 @@ foreach VSize = [2, 3, 4, 8, 16] in {
     }
   }
 }
-let MaxVersion = CL20 in {
+let Versions = [CL10, CL11, CL12, CL30] in {
   foreach AS = [GlobalAS, LocalAS, PrivateAS] in {
     def : Builtin<"vload_half", [Float, Size, PointerType<ConstType<Half>, AS>]>;
     foreach VSize = [2, 3, 4, 8, 16] in {
@@ -832,7 +843,8 @@ let MaxVersion = CL20 in {
     }
   }
 }
-let MinVersion = CL20 in {
+let Versions = [CL20, CL30] in {
+ let Extension = FuncExtGenericAddressSpace in {
   foreach AS = [GenericAS] in {
     def : Builtin<"vload_half", [Float, Size, PointerType<ConstType<Half>, AS>]>;
     foreach VSize = [2, 3, 4, 8, 16] in {
@@ -851,6 +863,7 @@ let MinVersion = CL20 in {
       }
     }
   }
+ }
 }
 
 foreach AS = [ConstantAS] in {
@@ -875,7 +888,9 @@ foreach name = ["async_work_group_strided_copy"] in {
   def : Builtin<name, [Event, PointerType<AGenTypeN, GlobalAS>, PointerType<ConstType<AGenTypeN>, LocalAS>, Size, Size, Event]>;
 }
 foreach name = ["wait_group_events"] in {
+ let Extension = FuncExtGenericAddressSpace in {
   def : Builtin<name, [Void, Int, PointerType<Event, GenericAS>]>;
+ }
 }
 foreach name = ["prefetch"] in {
   def : Builtin<name, [Void, PointerType<ConstType<AGenTypeN>, GlobalAS>, Size]>;
@@ -1154,7 +1169,8 @@ foreach aQual = ["WO", "RW"] in {
 //--------------------------------------------------------------------
 // OpenCL v2.0 s6.13.15 - Work-group Functions
 // --- Table 26 ---
-let MinVersion = CL20 in {
+let Versions = [CL20, CL30] in {
+ let Extension = FuncExtWorkGroupCollective in {
   foreach name = ["work_group_all", "work_group_any"] in {
     def : Builtin<name, [Int, Int], Attr.Convergent>;
   }
@@ -1169,11 +1185,12 @@ let MinVersion = CL20 in {
       def : Builtin<name # op, [IntLongFloatGenType1, IntLongFloatGenType1], Attr.Convergent>;
     }
   }
+ }
 }
 
 
 // OpenCL v2.0 s9.17.3: Additions to section 6.13.1: Work-Item Functions
-let MinVersion = CL20 in {
+let Versions = [CL20] in {
   let Extension = FuncExtKhrSubgroups in {
     def get_sub_group_size : Builtin<"get_sub_group_size", [UInt]>;
     def get_max_sub_group_size : Builtin<"get_max_sub_group_size", [UInt]>;
diff --git a/clang/lib/Sema/Sema.cpp b/clang/lib/Sema/Sema.cpp
index 9cfce5a63b1d..c26f45f62668 100644
--- a/clang/lib/Sema/Sema.cpp
+++ b/clang/lib/Sema/Sema.cpp
@@ -286,9 +286,10 @@ void Sema::Initialize() {
   // Initialize predefined OpenCL types and supported extensions and (optional)
   // core features.
   if (getLangOpts().OpenCL) {
+    getOpenCLOptions().setOpenCLVersion(getLangOpts());
     getOpenCLOptions().addSupport(
         Context.getTargetInfo().getSupportedOpenCLOpts());
-    getOpenCLOptions().enableSupportedCore(getLangOpts());
+    getOpenCLOptions().enableSupportedCore();
     addImplicitTypedef("sampler_t", Context.OCLSamplerTy);
     addImplicitTypedef("event_t", Context.OCLEventTy);
     if (getLangOpts().OpenCLCPlusPlus || getLangOpts().OpenCLVersion >= 200) {
@@ -344,12 +345,18 @@ void Sema::Initialize() {
             "cl_khr_int64_base_atomics cl_khr_int64_extended_atomics");
 
       setOpenCLExtensionForType(AtomicDoubleT, "cl_khr_fp64");
+      setOpenCLExtensionForType(Context.OCLReserveIDTy, "__opencl_c_pipes");
+      setOpenCLExtensionForType(Context.OCLClkEventTy,
+                                "__opencl_c_device_enqueue");
+      setOpenCLExtensionForType(Context.OCLQueueTy,
+                                "__opencl_c_device_enqueue");
     }
 
     setOpenCLExtensionForType(Context.DoubleTy, "cl_khr_fp64");
 
-#define GENERIC_IMAGE_TYPE_EXT(Type, Id, Ext) \
-    setOpenCLExtensionForType(Context.Id, Ext);
+#define GENERIC_IMAGE_TYPE_EXT(Type, Id, Ext)                                  \
+  setOpenCLExtensionForType(Context.Id, Ext);                                  \
+  setOpenCLExtensionForType(Context.Id, "__opencl_c_images");
 #include "clang/Basic/OpenCLImageTypes.def"
 #define EXT_OPAQUE_TYPE(ExtType, Id, Ext) \
     addImplicitTypedef(#ExtType, Context.Id##Ty); \
@@ -2293,6 +2300,27 @@ bool Sema::isOpenCLDisabledDecl(Decl *FD) {
   return false;
 }
 
+template <typename DiagLocT, typename DiagInfoT>
+void Sema::DiagnoseOpenCLRequiresOption(llvm::StringRef OpenCLOptName,
+                                        DiagLocT DiagLoc, DiagInfoT DiagInfo,
+                                        unsigned Selector,
+                                        SourceRange SrcRange) {
+  const auto &LO = getLangOpts();
+  auto CLVer = LO.OpenCLCPlusPlus ? 200 : LO.OpenCLVersion;
+  // For versions higher that 3.0 diagnosing feature
+  if (CLVer >= 300) {
+    OpenCLOptName =
+        llvm::StringSwitch<llvm::StringRef>(OpenCLOptName)
+            .Case("cl_khr_3d_image_writes", "__opencl_c_3d_image_writes")
+            .Case("cl_khr_subgroups", "__opencl_c_subgroups")
+            .Case("cl_khr_fp64", "__opencl_c_fp64")
+            .Default(OpenCLOptName);
+  }
+
+  Diag(DiagLoc, diag::err_opencl_requires_extension)
+      << Selector << DiagInfo << OpenCLOptName << SrcRange;
+}
+
 template <typename T, typename DiagLocT, typename DiagInfoT, typename MapT>
 bool Sema::checkOpenCLDisabledTypeOrDecl(T D, DiagLocT DiagLoc,
                                          DiagInfoT DiagInfo, MapT &Map,
@@ -2304,8 +2332,7 @@ bool Sema::checkOpenCLDisabledTypeOrDecl(T D, DiagLocT DiagLoc,
   bool Disabled = false;
   for (auto &I : Loc->second) {
     if (I != CurrOpenCLExtension && !getOpenCLOptions().isEnabled(I)) {
-      Diag(DiagLoc, diag::err_opencl_requires_extension) << Selector << DiagInfo
-                                                         << I << SrcRange;
+      DiagnoseOpenCLRequiresOption(I, DiagLoc, DiagInfo, Selector, SrcRange);
       Disabled = true;
     }
   }
@@ -2341,3 +2368,13 @@ bool Sema::checkOpenCLDisabledDecl(const NamedDecl &D, const Expr &E) {
   return checkOpenCLDisabledTypeOrDecl(&D, E.getBeginLoc(), FnName,
                                        OpenCLDeclExtMap, 1, D.getSourceRange());
 }
+
+bool Sema::checkOpenCLSubgroupExtForCallExpr(CallExpr *Call) {
+  if (!getOpenCLOptions().isEnabled("cl_khr_subgroups")) {
+    DiagnoseOpenCLRequiresOption("cl_khr_subgroups", Call->getBeginLoc(),
+                                 Call->getDirectCallee(), 1,
+                                 Call->getSourceRange());
+    return true;
+  }
+  return false;
+}
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index 74742023d1b3..efa3b6ab0eb6 100644
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -597,20 +597,11 @@ static bool checkOpenCLBlockArgs(Sema &S, Expr *BlockArg) {
   return IllegalParams;
 }
 
-static bool checkOpenCLSubgroupExt(Sema &S, CallExpr *Call) {
-  if (!S.getOpenCLOptions().isEnabled("cl_khr_subgroups")) {
-    S.Diag(Call->getBeginLoc(), diag::err_opencl_requires_extension)
-        << 1 << Call->getDirectCallee() << "cl_khr_subgroups";
-    return true;
-  }
-  return false;
-}
-
 static bool SemaOpenCLBuiltinNDRangeAndBlock(Sema &S, CallExpr *TheCall) {
   if (checkArgCount(S, TheCall, 2))
     return true;
 
-  if (checkOpenCLSubgroupExt(S, TheCall))
+  if (S.checkOpenCLSubgroupExtForCallExpr(TheCall))
     return true;
 
   // First argument is an ndrange_t type.
@@ -1564,7 +1555,7 @@ Sema::CheckBuiltinFunctionCall(FunctionDecl *FDecl, unsigned BuiltinID,
     break;
   case Builtin::BIsub_group_reserve_read_pipe:
   case Builtin::BIsub_group_reserve_write_pipe:
-    if (checkOpenCLSubgroupExt(*this, TheCall) ||
+    if (checkOpenCLSubgroupExtForCallExpr(TheCall) ||
         SemaBuiltinReserveRWPipe(*this, TheCall))
       return ExprError();
     break;
@@ -1577,7 +1568,7 @@ Sema::CheckBuiltinFunctionCall(FunctionDecl *FDecl, unsigned BuiltinID,
     break;
   case Builtin::BIsub_group_commit_read_pipe:
   case Builtin::BIsub_group_commit_write_pipe:
-    if (checkOpenCLSubgroupExt(*this, TheCall) ||
+    if (checkOpenCLSubgroupExtForCallExpr(TheCall) ||
         SemaBuiltinCommitRWPipe(*this, TheCall))
       return ExprError();
     break;
@@ -4314,6 +4305,20 @@ DiagnoseCStringFormatDirectiveInCFAPI(Sema &S,
   }
 }
 
+bool Sema::isSupportedOpenCLOMemoryOrdering(int64_t Ordering) const {
+  assert(llvm::isValidAtomicOrderingCABI(Ordering));
+  auto OrderingCABI = (llvm::AtomicOrderingCABI)Ordering;
+  switch (OrderingCABI) {
+  case llvm::AtomicOrderingCABI::acquire:
+  case llvm::AtomicOrderingCABI::release:
+  case llvm::AtomicOrderingCABI::acq_rel:
+    return OpenCLFeatures.isEnabled("__opencl_c_atomic_order_acq_rel");
+
+  default:
+    return true;
+  }
+}
+
 /// Determine whether the given type has a non-null nullability annotation.
 static bool isNonNullType(ASTContext &ctx, QualType type) {
   if (auto nullability = type->getNullability(ctx))
@@ -5067,10 +5072,17 @@ ExprResult Sema::BuildAtomicExpr(SourceRange CallRange, SourceRange ExprRange,
   if (SubExprs.size() >= 2 && Form != Init) {
     llvm::APSInt Result(32);
     if (SubExprs[1]->isIntegerConstantExpr(Result, Context) &&
-        !isValidOrderingForOp(Result.getSExtValue(), Op))
+        !isValidOrderingForOp(Result.getSExtValue(), Op)) {
       Diag(SubExprs[1]->getBeginLoc(),
            diag::warn_atomic_op_has_invalid_memory_order)
           << SubExprs[1]->getSourceRange();
+    } else if (IsOpenCL &&
+               !isSupportedOpenCLOMemoryOrdering(Result.getSExtValue())) {
+        Diag(SubExprs[1]->getBeginLoc(),
+             diag::err_opencl_memory_ordering_require_feat)
+            << SubExprs[1]->getSourceRange();
+        return ExprError();
+      }
   }
 
   if (auto ScopeModel = AtomicExpr::getScopeModel(Op)) {
diff --git a/clang/lib/Sema/SemaDecl.cpp b/clang/lib/Sema/SemaDecl.cpp
index 64146f4a912f..c1e629b5dc76 100644
--- a/clang/lib/Sema/SemaDecl.cpp
+++ b/clang/lib/Sema/SemaDecl.cpp
@@ -6152,7 +6152,9 @@ void Sema::deduceOpenCLAddressSpace(ValueDecl *Decl) {
     if (Type->isSamplerT() || Type->isVoidType())
       return;
     LangAS ImplAS = LangAS::opencl_private;
-    if ((getLangOpts().OpenCLCPlusPlus || getLangOpts().OpenCLVersion >= 200) &&
+    if ((getLangOpts().OpenCLCPlusPlus ||
+         getOpenCLOptions().isEnabled(
+             "__opencl_c_program_scope_global_variables")) &&
         Var->hasGlobalStorage())
       ImplAS = LangAS::opencl_global;
     // If the original type from a decayed type is an array type and that array
@@ -7682,6 +7684,10 @@ void Sema::CheckVariableDeclarationType(VarDecl *NewVD) {
     // OpenCL C v2.0 s6.5.1 - Variables defined at program scope and static
     // variables inside a function can also be declared in the global
     // address space.
+    // OpenCL C v3.0 s5.5 - For OpenCL C 2.0, or with the
+    // __opencl_c_program_scope_global_variables feature macro, the
+    // address space for a variable at program scope or a static variable
+    // also be __global
     // C++ for OpenCL inherits rule from OpenCL C v2.0.
     // FIXME: Adding local AS in C++ for OpenCL might make sense.
     if (NewVD->isFileVarDecl() || NewVD->isStaticLocal() ||
@@ -7689,10 +7695,11 @@ void Sema::CheckVariableDeclarationType(VarDecl *NewVD) {
       if (!T->isSamplerT() &&
           !(T.getAddressSpace() == LangAS::opencl_constant ||
             (T.getAddressSpace() == LangAS::opencl_global &&
-             (getLangOpts().OpenCLVersion == 200 ||
-              getLangOpts().OpenCLCPlusPlus)))) {
+             (OpenCLFeatures.isEnabled(
+                 "__opencl_c_program_scope_global_variables"))))) {
         int Scope = NewVD->isStaticLocal() | NewVD->hasExternalStorage() << 1;
-        if (getLangOpts().OpenCLVersion == 200 || getLangOpts().OpenCLCPlusPlus)
+        if (OpenCLFeatures.isEnabled(
+                "__opencl_c_program_scope_global_variables"))
           Diag(NewVD->getLocation(), diag::err_opencl_global_invalid_addr_space)
               << Scope << "global or constant";
         else
diff --git a/clang/lib/Sema/SemaDeclAttr.cpp b/clang/lib/Sema/SemaDeclAttr.cpp
index 849bc09063b3..c963b95b131a 100644
--- a/clang/lib/Sema/SemaDeclAttr.cpp
+++ b/clang/lib/Sema/SemaDeclAttr.cpp
@@ -6362,7 +6362,7 @@ static void handleInternalLinkageAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
 }
 
 static void handleOpenCLNoSVMAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
-  if (S.LangOpts.OpenCLVersion != 200)
+  if (S.LangOpts.OpenCLVersion < 200)
     S.Diag(AL.getLoc(), diag::err_attribute_requires_opencl_version)
         << AL << "2.0" << 0;
   else
@@ -6446,6 +6446,13 @@ static void handleOpenCLAccessAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
             << AL << PDecl->getType() << DeclTy->isImageType();
         D->setInvalidDecl(true);
         return;
+      } else if ((!S.getLangOpts().OpenCLCPlusPlus &&
+                  S.getLangOpts().OpenCLVersion >= 200) &&
+                 !S.getOpenCLOptions().isEnabled(
+                     "__opencl_c_read_write_images")) {
+        S.Diag(AL.getLoc(), diag::err_opencl_requires_extension)
+            << 0 << PDecl->getType() << "__opencl_c_read_write_images";
+        return;
       }
     }
   }
diff --git a/clang/lib/Sema/SemaDeclCXX.cpp b/clang/lib/Sema/SemaDeclCXX.cpp
index 831e55046e80..4481a59b4517 100644
--- a/clang/lib/Sema/SemaDeclCXX.cpp
+++ b/clang/lib/Sema/SemaDeclCXX.cpp
@@ -14906,6 +14906,11 @@ CheckOperatorNewDeleteTypes(Sema &SemaRef, const FunctionDecl *FnDecl,
     if (auto *PtrTy = ResultType->getAs<PointerType>()) {
       ResultType = RemoveAddressSpaceFromPtr(SemaRef, PtrTy);
     }
+    if (CanQual<PointerType> ExpectedPtrTy =
+            ExpectedResultType->getAs<PointerType>()) {
+      ExpectedResultType = SemaRef.Context.getCanonicalType(
+          RemoveAddressSpaceFromPtr(SemaRef, ExpectedPtrTy->getTypePtr()));
+    }
   }
 
   // Check that the result type is what we expect.
@@ -14939,6 +14944,11 @@ CheckOperatorNewDeleteTypes(Sema &SemaRef, const FunctionDecl *FnDecl,
             FnDecl->getParamDecl(0)->getType()->getAs<PointerType>()) {
       FirstParamType = RemoveAddressSpaceFromPtr(SemaRef, PtrTy);
     }
+    if (CanQual<PointerType> ExpectedPtrTy =
+            ExpectedFirstParamType->getAs<PointerType>()) {
+      ExpectedFirstParamType = SemaRef.Context.getCanonicalType(
+          RemoveAddressSpaceFromPtr(SemaRef, ExpectedPtrTy->getTypePtr()));
+    }
   }
   if (SemaRef.Context.getCanonicalType(FirstParamType).getUnqualifiedType() !=
       ExpectedFirstParamType)
diff --git a/clang/lib/Sema/SemaLookup.cpp b/clang/lib/Sema/SemaLookup.cpp
index 8d96404a5c27..8f21203b9508 100644
--- a/clang/lib/Sema/SemaLookup.cpp
+++ b/clang/lib/Sema/SemaLookup.cpp
@@ -771,19 +771,20 @@ static void InsertOCLBuiltinDeclarationsFromTable(Sema &S, LookupResult &LR,
   // as argument.  Only meaningful for generic types, otherwise equals 1.
   unsigned GenTypeMaxCnt;
 
+  ASTContext &Context = S.Context;
+
+  // Ignore this BIF if its version does not match the language options.
+  unsigned OpenCLVersion = Context.getLangOpts().OpenCLVersion;
+  if (Context.getLangOpts().OpenCLCPlusPlus)
+    OpenCLVersion = 200;
+
+  unsigned short VersionCode = OpenCLOptions::EncodeVersion(OpenCLVersion);
+
   for (unsigned SignatureIndex = 0; SignatureIndex < Len; SignatureIndex++) {
     const OpenCLBuiltinStruct &OpenCLBuiltin =
         BuiltinTable[FctIndex + SignatureIndex];
-    ASTContext &Context = S.Context;
 
-    // Ignore this BIF if its version does not match the language options.
-    unsigned OpenCLVersion = Context.getLangOpts().OpenCLVersion;
-    if (Context.getLangOpts().OpenCLCPlusPlus)
-      OpenCLVersion = 200;
-    if (OpenCLVersion < OpenCLBuiltin.MinVersion)
-      continue;
-    if ((OpenCLBuiltin.MaxVersion != 0) &&
-        (OpenCLVersion >= OpenCLBuiltin.MaxVersion))
+    if (!(OpenCLBuiltin.AllVersions & VersionCode))
       continue;
 
     SmallVector<QualType, 1> RetTypes;
diff --git a/clang/lib/Sema/SemaType.cpp b/clang/lib/Sema/SemaType.cpp
index 93ddd047e09b..93211b801f8d 100644
--- a/clang/lib/Sema/SemaType.cpp
+++ b/clang/lib/Sema/SemaType.cpp
@@ -2016,7 +2016,7 @@ static QualType deduceOpenCLPointeeAddrSpace(Sema &S, QualType PointeeType) {
       !PointeeType.hasAddressSpace())
     PointeeType = S.getASTContext().getAddrSpaceQualType(
         PointeeType,
-        S.getLangOpts().OpenCLCPlusPlus || S.getLangOpts().OpenCLVersion == 200
+        S.getOpenCLOptions().isEnabled("__opencl_c_generic_address_space")
             ? LangAS::opencl_generic
             : LangAS::opencl_private);
   return PointeeType;
@@ -5160,9 +5160,15 @@ static TypeSourceInfo *GetFullTypeForDeclarator(TypeProcessingState &state,
     }
 
     case DeclaratorChunk::Pipe: {
-      T = S.BuildReadPipeType(T, DeclType.Loc);
-      processTypeAttrs(state, T, TAL_DeclSpec,
-                       D.getMutableDeclSpec().getAttributes());
+      if (S.getOpenCLOptions().isEnabled("__opencl_c_pipes")) {
+        T = S.BuildReadPipeType(T, DeclType.Loc);
+        processTypeAttrs(state, T, TAL_DeclSpec,
+                         D.getMutableDeclSpec().getAttributes());
+      } else {
+        D.setInvalidType(true);
+        T = Context.IntTy;
+        S.Diag(D.getIdentifierLoc(), diag::err_opencl_pipes_require_feat);
+      }
       break;
     }
     }
diff --git a/clang/test/CodeGenOpenCL/addr-space-struct-arg.cl b/clang/test/CodeGenOpenCL/addr-space-struct-arg.cl
index cdbf28bbcad8..0bedff5ef0f3 100644
--- a/clang/test/CodeGenOpenCL/addr-space-struct-arg.cl
+++ b/clang/test/CodeGenOpenCL/addr-space-struct-arg.cl
@@ -2,7 +2,8 @@
 // RUN: %clang_cc1 %s -emit-llvm -o - -O0 -triple amdgcn | FileCheck -enable-var-scope -check-prefixes=COM,AMDGCN %s
 // RUN: %clang_cc1 %s -emit-llvm -o - -cl-std=CL2.0 -O0 -triple amdgcn | FileCheck -enable-var-scope -check-prefixes=COM,AMDGCN,AMDGCN20 %s
 // RUN: %clang_cc1 %s -emit-llvm -o - -cl-std=CL1.2 -O0 -triple spir-unknown-unknown-unknown | FileCheck -enable-var-scope -check-prefixes=SPIR %s
-
+// RUN: %clang_cc1 %s -emit-llvm -o - -cl-std=CL3.0 -O0 -triple spir-unknown-unknown-unknown | FileCheck -enable-var-scope -check-prefixes=SPIR %s
+// RUN: %clang_cc1 %s -emit-llvm -o - -cl-std=CL3.0 -cl-ext=__opencl_c_program_scope_global_variables -O0 -triple spir-unknown-unknown-unknown | FileCheck -enable-var-scope -check-prefixes=SPIR %s
 typedef int int2 __attribute__((ext_vector_type(2)));
 
 typedef struct {
@@ -39,7 +40,7 @@ struct LargeStructTwoMember {
   int2 y[20];
 };
 
-#if __OPENCL_C_VERSION__ >= 200
+#ifdef __opencl_c_program_scope_global_variables
 struct LargeStructOneMember g_s;
 #endif
 
@@ -98,7 +99,7 @@ void FuncOneLargeMember(struct LargeStructOneMember u) {
 // AMDGCN20:  %[[r0:.*]] = bitcast %struct.LargeStructOneMember addrspace(5)* %[[byval_temp]] to i8 addrspace(5)*
 // AMDGCN20:  call void @llvm.memcpy.p5i8.p1i8.i64(i8 addrspace(5)* align 8 %[[r0]], i8 addrspace(1)* align 8 bitcast (%struct.LargeStructOneMember addrspace(1)* @g_s to i8 addrspace(1)*), i64 800, i1 false)
 // AMDGCN20:  call void @FuncOneLargeMember(%struct.LargeStructOneMember addrspace(5)* byval(%struct.LargeStructOneMember) align 8 %[[byval_temp]])
-#if __OPENCL_C_VERSION__ >= 200
+#ifdef __opencl_c_program_scope_global_variables
 void test_indirect_arg_globl(void) {
   FuncOneLargeMember(g_s);
 }
diff --git a/clang/test/CodeGenOpenCL/address-spaces-conversions.cl b/clang/test/CodeGenOpenCL/address-spaces-conversions.cl
index 52feccc540bb..9ecffcca5ee9 100644
--- a/clang/test/CodeGenOpenCL/address-spaces-conversions.cl
+++ b/clang/test/CodeGenOpenCL/address-spaces-conversions.cl
@@ -1,5 +1,7 @@
 // RUN: %clang_cc1 %s -triple x86_64-unknown-linux-gnu -O0 -ffake-address-space-map -cl-std=CL2.0 -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 %s -triple x86_64-unknown-linux-gnu -O0 -ffake-address-space-map -cl-std=CL3.0 -cl-ext=__opencl_c_generic_address_space -emit-llvm -o - | FileCheck %s
 // RUN: %clang_cc1 %s -triple x86_64-unknown-linux-gnu -O0 -cl-std=CL2.0 -emit-llvm -o - | FileCheck --check-prefix=CHECK-NOFAKE %s
+// RUN: %clang_cc1 %s -triple x86_64-unknown-linux-gnu -O0 -cl-std=CL3.0 -cl-ext=__opencl_c_generic_address_space -emit-llvm -o - | FileCheck --check-prefix=CHECK-NOFAKE %s
 // When -ffake-address-space-map is not used, all addr space mapped to 0 for x86_64.
 
 // test that we generate address space casts everywhere we need conversions of
diff --git a/clang/test/CodeGenOpenCL/address-spaces-mangling.cl b/clang/test/CodeGenOpenCL/address-spaces-mangling.cl
index 50622f099143..e19ec8451d0d 100644
--- a/clang/test/CodeGenOpenCL/address-spaces-mangling.cl
+++ b/clang/test/CodeGenOpenCL/address-spaces-mangling.cl
@@ -6,6 +6,7 @@
 // We check that the address spaces are mangled the same in both version of OpenCL
 // RUN: %clang_cc1 %s -triple spir-unknown-unknown -cl-std=CL2.0 -emit-llvm -o - | FileCheck -check-prefix=OCL-20 %s
 // RUN: %clang_cc1 %s -triple spir-unknown-unknown -cl-std=CL1.2 -emit-llvm -o - | FileCheck -check-prefix=OCL-12 %s
+// RUN: %clang_cc1 %s -triple spir-unknown-unknown -cl-std=CL3.0 -emit-llvm -o - | FileCheck -check-prefix=OCL-12 %s
 
 // We can't name this f as private is equivalent to default
 // no specifier given address space so we get multiple definition
@@ -47,7 +48,7 @@ void f(constant int *arg) { }
 // OCL-20-DAG: @_Z1fPU3AS2i
 // OCL-12-DAG: @_Z1fPU3AS2i
 
-#if __OPENCL_C_VERSION__ >= 200
+#if __OPENCL_C_VERSION__ == 200
 __attribute__((overloadable))
 void f(generic int *arg) { }
 // ASMANG20: @_Z1fPU3AS4i
diff --git a/clang/test/CodeGenOpenCL/address-spaces.cl b/clang/test/CodeGenOpenCL/address-spaces.cl
index 3c8fea2a80bc..26a741338b3a 100644
--- a/clang/test/CodeGenOpenCL/address-spaces.cl
+++ b/clang/test/CodeGenOpenCL/address-spaces.cl
@@ -1,9 +1,13 @@
 // RUN: %clang_cc1 %s -O0 -ffake-address-space-map -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK,SPIR
+// RUN: %clang_cc1 %s -O0 -cl-std=CL3.0 -ffake-address-space-map -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK,SPIR
 // RUN: %clang_cc1 %s -O0 -DCL20 -cl-std=CL2.0 -ffake-address-space-map -emit-llvm -o - | FileCheck %s --check-prefixes=CL20,CL20SPIR
 // RUN: %clang_cc1 %s -O0 -triple amdgcn-amd-amdhsa -emit-llvm -o - | FileCheck --check-prefixes=CHECK,AMDGCN %s
+// RUN: %clang_cc1 %s -O0 -cl-std=CL3.0 -triple amdgcn-amd-amdhsa -emit-llvm -o - | FileCheck --check-prefixes=CHECK,AMDGCN %s
 // RUN: %clang_cc1 %s -O0 -triple amdgcn-amd-amdhsa -DCL20 -cl-std=CL2.0 -emit-llvm -o - | FileCheck %s --check-prefixes=CL20,CL20AMDGCN
 // RUN: %clang_cc1 %s -O0 -triple amdgcn-mesa-mesa3d -emit-llvm -o - | FileCheck --check-prefixes=CHECK,AMDGCN %s
+// RUN: %clang_cc1 %s -O0 -cl-std=CL3.0 -triple amdgcn-mesa-mesa3d -emit-llvm -o - | FileCheck --check-prefixes=CHECK,AMDGCN %s
 // RUN: %clang_cc1 %s -O0 -triple r600-- -emit-llvm -o - | FileCheck --check-prefixes=CHECK,AMDGCN %s
+// RUN: %clang_cc1 %s -O0 -cl-std=CL3.0 -triple r600-- -emit-llvm -o - | FileCheck --check-prefixes=CHECK,AMDGCN %s
 
 // SPIR: %struct.S = type { i32, i32, i32* }
 // CL20SPIR: %struct.S = type { i32, i32, i32 addrspace(4)* }
diff --git a/clang/test/CodeGenOpenCL/amdgcn-automatic-variable.cl b/clang/test/CodeGenOpenCL/amdgcn-automatic-variable.cl
index 7216cb517495..8d3b30fe8074 100644
--- a/clang/test/CodeGenOpenCL/amdgcn-automatic-variable.cl
+++ b/clang/test/CodeGenOpenCL/amdgcn-automatic-variable.cl
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 -O0 -cl-std=CL1.2 -triple amdgcn---amdgizcl -emit-llvm %s -o - | FileCheck -check-prefixes=CHECK,CL12 %s
+// RUN: %clang_cc1 -O0 -cl-std=CL3.0 -triple amdgcn---amdgizcl -emit-llvm %s -o - | FileCheck -check-prefixes=CHECK,CL12 %s
 // RUN: %clang_cc1 -O0 -cl-std=CL2.0 -triple amdgcn---amdgizcl -emit-llvm %s -o - | FileCheck -check-prefixes=CHECK,CL20 %s
 
 // CL12-LABEL: define void @func1(i32 addrspace(5)* %x)
diff --git a/clang/test/CodeGenOpenCL/amdgpu-sizeof-alignof.cl b/clang/test/CodeGenOpenCL/amdgpu-sizeof-alignof.cl
index a5d438933fa4..8073c7756eb6 100644
--- a/clang/test/CodeGenOpenCL/amdgpu-sizeof-alignof.cl
+++ b/clang/test/CodeGenOpenCL/amdgpu-sizeof-alignof.cl
@@ -4,6 +4,17 @@
 // RUN: %clang_cc1 -triple amdgcn---opencl -cl-std=CL2.0 %s -emit-llvm -o - | FileCheck %s
 // RUN: %clang_cc1 -triple amdgcn---amdgizcl -cl-std=CL1.2 %s -emit-llvm -o - | FileCheck %s
 // RUN: %clang_cc1 -triple amdgcn---amdgizcl -cl-std=CL2.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple r600 -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn-mesa-mesa3d -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn---opencl -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn---amdgizcl -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn-mesa-mesa3d -cl-ext=__opencl_c_generic_address_space -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn---opencl -cl-ext=__opencl_c_generic_address_space -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn---amdgizcl -cl-ext=__opencl_c_generic_address_space -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple r600 -cl-ext=__opencl_c_fp64 -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn-mesa-mesa3d -cl-ext=__opencl_c_fp64 -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn---opencl -cl-ext=__opencl_c_fp64 -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn---amdgizcl -cl-ext=__opencl_c_fp64 -cl-std=CL3.0 %s -emit-llvm -o - | FileCheck %s
 
 #ifdef __AMDGCN__
 #define PTSIZE 8
@@ -11,7 +22,7 @@
 #define PTSIZE 4
 #endif
 
-#ifdef cl_khr_fp64
+#if defined(cl_khr_fp64) || defined(__opencl_c_fp64)
 #pragma OPENCL EXTENSION cl_khr_fp64 : enable
 #endif
 #ifdef cl_khr_fp16
@@ -59,8 +70,12 @@ void test() {
   check(__alignof__(double) == 8);
 #endif
 
-  check(sizeof(void*) == (__OPENCL_C_VERSION__ >= 200 ? 8 : 4));
-  check(__alignof__(void*) == (__OPENCL_C_VERSION__ >= 200 ? 8 : 4));
+  check(sizeof(private void*) == 4);
+  check(__alignof__(private void*) == 4);
+#ifdef __opencl_c_generic_address_space
+  check(sizeof(generic void*) == 8);
+  check(__alignof__(generic void*) == 8);
+#endif
   check(sizeof(global_ptr_t) == PTSIZE);
   check(__alignof__(global_ptr_t) == PTSIZE);
   check(sizeof(constant_ptr_t) == PTSIZE);
diff --git a/clang/test/CodeGenOpenCL/arm-integer-dot-product.cl b/clang/test/CodeGenOpenCL/arm-integer-dot-product.cl
index d1ab6aceac5c..70c5bace023b 100644
--- a/clang/test/CodeGenOpenCL/arm-integer-dot-product.cl
+++ b/clang/test/CodeGenOpenCL/arm-integer-dot-product.cl
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 %s -triple spir-unknown-unknown -finclude-default-header -cl-std=CL1.2 -emit-llvm -o - -O0 | FileCheck %s
+// RUN: %clang_cc1 %s -triple spir-unknown-unknown -finclude-default-header -cl-std=CL3.0 -emit-llvm -o - -O0 | FileCheck %s
 
 #pragma OPENCL EXTENSION cl_arm_integer_dot_product_int8 : enable
 void test_int8(uchar4 ua, uchar4 ub, char4 sa, char4 sb) {
diff --git a/clang/test/CodeGenOpenCL/cl-uniform-wg-size.cl b/clang/test/CodeGenOpenCL/cl-uniform-wg-size.cl
index 76ace5dca21e..5dc43e222f75 100644
--- a/clang/test/CodeGenOpenCL/cl-uniform-wg-size.cl
+++ b/clang/test/CodeGenOpenCL/cl-uniform-wg-size.cl
@@ -1,6 +1,8 @@
 // RUN: %clang_cc1 -emit-llvm -O0 -cl-std=CL1.2 -o - %s 2>&1 | FileCheck %s -check-prefixes CHECK,CHECK-UNIFORM
 // RUN: %clang_cc1 -emit-llvm -O0 -cl-std=CL2.0 -o - %s 2>&1 | FileCheck %s -check-prefixes CHECK,CHECK-NONUNIFORM
 // RUN: %clang_cc1 -emit-llvm -O0 -cl-std=CL2.0 -cl-uniform-work-group-size -o - %s 2>&1 | FileCheck %s -check-prefixes CHECK,CHECK-UNIFORM
+// RUN: %clang_cc1 -emit-llvm -O0 -cl-std=CL3.0 -o - %s 2>&1 | FileCheck %s -check-prefixes CHECK,CHECK-NONUNIFORM
+// RUN: %clang_cc1 -emit-llvm -O0 -cl-std=CL3.0 -cl-uniform-work-group-size -o - %s 2>&1 | FileCheck %s -check-prefixes CHECK,CHECK-UNIFORM
 
 kernel void ker() {};
 // CHECK: define{{.*}}@ker() #0
diff --git a/clang/test/CodeGenOpenCL/fpmath.cl b/clang/test/CodeGenOpenCL/fpmath.cl
index 0108d909c94e..b28392739c71 100644
--- a/clang/test/CodeGenOpenCL/fpmath.cl
+++ b/clang/test/CodeGenOpenCL/fpmath.cl
@@ -2,6 +2,8 @@
 // RUN: %clang_cc1 %s -emit-llvm -o - -triple spir-unknown-unknown -cl-fp32-correctly-rounded-divide-sqrt | FileCheck --check-prefix=CHECK --check-prefix=DIVOPT %s
 // RUN: %clang_cc1 %s -emit-llvm -o - -DNOFP64 -cl-std=CL1.2 -triple r600-unknown-unknown -target-cpu r600 -pedantic | FileCheck --check-prefix=CHECK-FLT %s
 // RUN: %clang_cc1 %s -emit-llvm -o - -DFP64 -cl-std=CL1.2 -triple spir-unknown-unknown -pedantic | FileCheck --check-prefix=CHECK-DBL %s
+// RUN: %clang_cc1 %s -emit-llvm -o - -DNOFP64 -cl-std=CL3.0 -triple r600-unknown-unknown -target-cpu r600 -pedantic | FileCheck --check-prefix=CHECK-FLT %s
+// RUN: %clang_cc1 %s -emit-llvm -o - -DFP64 -cl-std=CL3.0 -cl-ext=__opencl_c_fp64 -triple spir-unknown-unknown -pedantic | FileCheck --check-prefix=CHECK-DBL %s
 
 typedef __attribute__(( ext_vector_type(4) )) float float4;
 
diff --git a/clang/test/CodeGenOpenCL/generic-address-space-feature.cl b/clang/test/CodeGenOpenCL/generic-address-space-feature.cl
new file mode 100644
index 000000000000..890860abe4d9
--- /dev/null
+++ b/clang/test/CodeGenOpenCL/generic-address-space-feature.cl
@@ -0,0 +1,28 @@
+// RUN: %clang_cc1 %s -triple  spir-unknown-unknown -O0 -cl-std=CL1.2 -cl-ext=-cl_khr_fp64 -finclude-default-header -emit-llvm -o - | FileCheck %s --check-prefixes=CL12
+// RUN: %clang_cc1 %s -triple  spir-unknown-unknown -O0 -cl-std=CL1.2 -cl-ext=-cl_khr_fp64 -fdeclare-opencl-builtins -emit-llvm -o - | FileCheck %s --check-prefixes=CL12
+// RUN: %clang_cc1 %s -triple  spir-unknown-unknown -O0 -cl-std=CL2.0 -cl-ext=-cl_khr_fp64 -finclude-default-header -emit-llvm -o - | FileCheck %s --check-prefixes=CL20
+// RUN: %clang_cc1 %s -triple  spir-unknown-unknown -O0 -cl-std=CL2.0 -cl-ext=-cl_khr_fp64 -fdeclare-opencl-builtins -emit-llvm -o - | FileCheck %s --check-prefixes=CL20
+// RUN: %clang_cc1 %s -triple  spir-unknown-unknown -O0 -cl-std=CL3.0 -cl-ext=-cl_khr_fp64 -finclude-default-header -emit-llvm -o - | FileCheck %s --check-prefixes=CL30
+// RUN: %clang_cc1 %s -triple  spir-unknown-unknown -O0 -cl-std=CL3.0 -cl-ext=-cl_khr_fp64,__opencl_c_generic_address_space -finclude-default-header -emit-llvm -o - | FileCheck %s --check-prefixes=CL30-GENERIC
+// RUN: %clang_cc1 %s -triple  spir-unknown-unknown -O0 -cl-std=CL3.0 -cl-ext=-cl_khr_fp64 -fdeclare-opencl-builtins -emit-llvm -o - | FileCheck %s --check-prefixes=CL30
+// RUN: %clang_cc1 %s -triple  spir-unknown-unknown -O0 -cl-std=CL3.0 -cl-ext=-cl_khr_fp64,__opencl_c_generic_address_space -fdeclare-opencl-builtins -emit-llvm -o - | FileCheck %s --check-prefixes=CL30-GENERIC
+
+void test(global float* src1, local float *src2, private float *src3, float *src4, float tmp) {
+    // CL20: %{{.+}} = addrspacecast float addrspace(1)* %{{.+}} to float addrspace(4)*
+    // CL12-NOT: addrspacecast
+    // CL30-NOT: addrspacecast
+    // CL30-GENERIC-NOT: addrspacecast
+    tmp = sincos(tmp, src1);
+    // CL20: %{{.+}} = addrspacecast float addrspace(3)* %{{.+}} to float addrspace(4)*
+    // CL12-NOT: addrspacecast
+    // CL30-NOT: addrspacecast
+    // CL30-GENERIC-NOT: addrspacecast
+    tmp = sincos(tmp, src2);
+
+    // CL12: %{{.+}} = call {{(spir_func )?}}float [[FUNCNAME:@.*sincos.*]](float %{{.+}}, float* {{.+}})
+    // CL20: %{{.+}} = call {{(spir_func )?}}float [[FUNCNAME:@.*sincos.*]](float %{{.+}}, float addrspace(4)* {{.+}})
+    // CL30: %{{.+}} = call {{(spir_func )?}}float [[FUNCNAME:@.*sincos.*]](float %{{.+}}, float* {{.+}})
+    // CL30-GENERIC: %{{.+}} = call {{(spir_func )?}}float [[FUNCNAME:@.*sincos.*]](float %{{.+}}, float addrspace(4)* {{.+}})
+    // CHECK: addrspacecast
+    tmp = sincos(tmp, src4);
+}
diff --git a/clang/test/CodeGenOpenCL/intel-subgroups-avc-ext-types.cl b/clang/test/CodeGenOpenCL/intel-subgroups-avc-ext-types.cl
index 515f13f6e768..5aa31ac6f345 100644
--- a/clang/test/CodeGenOpenCL/intel-subgroups-avc-ext-types.cl
+++ b/clang/test/CodeGenOpenCL/intel-subgroups-avc-ext-types.cl
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 %s -triple spir-unknown-unknown -cl-std=CL1.2 -cl-ext=+cl_intel_device_side_avc_motion_estimation -emit-llvm -o - -O0 | FileCheck %s
+// RUN: %clang_cc1 %s -triple spir-unknown-unknown -cl-std=CL3.0 -cl-ext=+cl_intel_device_side_avc_motion_estimation -emit-llvm -o - -O0 | FileCheck %s
 
 // CHECK: %opencl.intel_sub_group_avc_mce_payload_t = type opaque
 // CHECK: %opencl.intel_sub_group_avc_ime_payload_t = type opaque
diff --git a/clang/test/CodeGenOpenCL/kernels-have-spir-cc-by-default.cl b/clang/test/CodeGenOpenCL/kernels-have-spir-cc-by-default.cl
index e89237623478..3d6708ac361f 100644
--- a/clang/test/CodeGenOpenCL/kernels-have-spir-cc-by-default.cl
+++ b/clang/test/CodeGenOpenCL/kernels-have-spir-cc-by-default.cl
@@ -1,5 +1,8 @@
 // RUN: %clang_cc1 %s -cl-std=CL1.2 -emit-llvm -triple x86_64-unknown-unknown -o - | FileCheck %s
 // RUN: %clang_cc1 %s -cl-std=CL1.2 -emit-llvm -triple amdgcn-unknown-unknown -o - | FileCheck -check-prefixes=AMDGCN %s
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -cl-ext=__opencl_c_fp64 -emit-llvm -triple x86_64-unknown-unknown -o - | FileCheck %s
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -cl-ext=__opencl_c_fp64 -emit-llvm -triple amdgcn-unknown-unknown -o - | FileCheck -check-prefixes=AMDGCN %s
+
 // Test that the kernels always use the SPIR calling convention
 // to have unambiguous mapping of arguments to feasibly implement
 // clSetKernelArg().
diff --git a/clang/test/CodeGenOpenCL/logical-ops.cl b/clang/test/CodeGenOpenCL/logical-ops.cl
index f083a8580ee7..499eab65039b 100644
--- a/clang/test/CodeGenOpenCL/logical-ops.cl
+++ b/clang/test/CodeGenOpenCL/logical-ops.cl
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 %s -emit-llvm -o - -cl-std=CL1.2 -O1 -triple x86_64-unknown-linux-gnu | FileCheck %s
+// RUN: %clang_cc1 %s -emit-llvm -o - -cl-std=CL3.0 -cl-ext=__opencl_c_fp64 -O1 -triple x86_64-unknown-linux-gnu | FileCheck %s
 // RUN: %clang_cc1 %s -emit-llvm -o - -cl-std=clc++ -O1 -triple x86_64-unknown-linux-gnu | FileCheck %s
 
 #pragma OPENCL EXTENSION cl_khr_fp64 : enable
diff --git a/clang/test/CodeGenOpenCL/no-half.cl b/clang/test/CodeGenOpenCL/no-half.cl
index aee8f678f01a..46da7fa339e8 100644
--- a/clang/test/CodeGenOpenCL/no-half.cl
+++ b/clang/test/CodeGenOpenCL/no-half.cl
@@ -1,6 +1,7 @@
 // RUN: %clang_cc1 %s -cl-std=cl2.0 -emit-llvm -o - -triple spir-unknown-unknown | FileCheck %s
 // RUN: %clang_cc1 %s -cl-std=cl1.2 -emit-llvm -o - -triple spir-unknown-unknown | FileCheck %s
 // RUN: %clang_cc1 %s -cl-std=cl1.1 -emit-llvm -o - -triple spir-unknown-unknown | FileCheck %s
+// RUN: %clang_cc1 %s -cl-std=cl3.0 -cl-ext=__opencl_c_fp64 -emit-llvm -o - -triple spir-unknown-unknown | FileCheck %s
 
 #pragma OPENCL EXTENSION cl_khr_fp64:enable
 
diff --git a/clang/test/CodeGenOpenCL/pipe_builtin.cl b/clang/test/CodeGenOpenCL/pipe_builtin.cl
index 02b9669b7ab1..0bf35c336199 100644
--- a/clang/test/CodeGenOpenCL/pipe_builtin.cl
+++ b/clang/test/CodeGenOpenCL/pipe_builtin.cl
@@ -1,4 +1,7 @@
 // RUN: %clang_cc1 -triple %itanium_abi_triple -emit-llvm -cl-ext=+cl_khr_subgroups -O0 -cl-std=clc++ -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple %itanium_abi_triple -emit-llvm -cl-ext=+cl_khr_subgroups -O0 -cl-std=cl2.0 -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple %itanium_abi_triple -emit-llvm -cl-ext=+cl_khr_subgroups -O0 -cl-std=cl3.0 -cl-ext=__opencl_c_pipes,__opencl_c_subgroups -o - %s | FileCheck %s
+
 // FIXME: Add MS ABI manglings of OpenCL things and remove %itanium_abi_triple
 // above to support OpenCL in the MS C++ ABI.
 
diff --git a/clang/test/CodeGenOpenCL/pipe_types.cl b/clang/test/CodeGenOpenCL/pipe_types.cl
index ba064c6d7557..b7a523d4f084 100644
--- a/clang/test/CodeGenOpenCL/pipe_types.cl
+++ b/clang/test/CodeGenOpenCL/pipe_types.cl
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -O0 -cl-std=CL2.0 -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -O0 -cl-std=CL3.0 -cl-ext=__opencl_c_pipes -o - %s | FileCheck %s
 
 // CHECK: %opencl.pipe_ro_t = type opaque
 // CHECK: %opencl.pipe_wo_t = type opaque
diff --git a/clang/test/CodeGenOpenCL/printf.cl b/clang/test/CodeGenOpenCL/printf.cl
index fc139d776db6..0133c5595d81 100644
--- a/clang/test/CodeGenOpenCL/printf.cl
+++ b/clang/test/CodeGenOpenCL/printf.cl
@@ -1,5 +1,7 @@
 // RUN: %clang_cc1 -cl-std=CL1.2 -cl-ext=-+cl_khr_fp64 -triple spir-unknown-unknown -disable-llvm-passes -emit-llvm -o - %s | FileCheck -check-prefixes=FP64,ALL %s
 // RUN: %clang_cc1 -cl-std=CL1.2 -cl-ext=-cl_khr_fp64 -triple spir-unknown-unknown -disable-llvm-passes -emit-llvm -o - %s | FileCheck -check-prefixes=NOFP64,ALL %s
+// RUN: %clang_cc1 -cl-std=CL3.0 -cl-ext=+__opencl_c_fp64 -triple spir-unknown-unknown -disable-llvm-passes -emit-llvm -o - %s | FileCheck -check-prefixes=FP64,ALL %s
+// RUN: %clang_cc1 -cl-std=CL3.0 -triple spir-unknown-unknown -disable-llvm-passes -emit-llvm -o - %s | FileCheck -check-prefixes=NOFP64,ALL %s
 
 typedef __attribute__((ext_vector_type(2))) float float2;
 typedef __attribute__((ext_vector_type(2))) half half2;
diff --git a/clang/test/CodeGenOpenCL/unroll-hint.cl b/clang/test/CodeGenOpenCL/unroll-hint.cl
index 0f84450a1ae6..9347c935869b 100644
--- a/clang/test/CodeGenOpenCL/unroll-hint.cl
+++ b/clang/test/CodeGenOpenCL/unroll-hint.cl
@@ -1,5 +1,6 @@
 // RUN: %clang_cc1 -emit-llvm -O0 -cl-std=CL2.0 -o - %s | FileCheck %s
 // RUN: %clang_cc1 -emit-llvm -O0 -cl-std=CL1.2 -o - %s | FileCheck %s
+// RUN: %clang_cc1 -emit-llvm -O0 -cl-std=CL3.0 -o - %s | FileCheck %s
 
 /*** for ***/
 void for_count()
diff --git a/clang/test/Driver/autocomplete.c b/clang/test/Driver/autocomplete.c
index 5c0bfb69f9a3..eabdf0b5938d 100644
--- a/clang/test/Driver/autocomplete.c
+++ b/clang/test/Driver/autocomplete.c
@@ -43,6 +43,8 @@
 // CLSTDALL-NEXT: CL1.2
 // CLSTDALL-NEXT: cl2.0
 // CLSTDALL-NEXT: CL2.0
+// CLSTDALL-NEXT: cl3.0
+// CLSTDALL-NEXT: CL3.0
 // CLSTDALL-NEXT: clc++
 // CLSTDALL-NEXT: CLC++
 // RUN: %clang --autocomplete=-fno-sanitize-coverage=,f | FileCheck %s -check-prefix=FNOSANICOVER
diff --git a/clang/test/Driver/opencl.cl b/clang/test/Driver/opencl.cl
index 63b04bc1af41..ffdd4f7d65bb 100644
--- a/clang/test/Driver/opencl.cl
+++ b/clang/test/Driver/opencl.cl
@@ -2,6 +2,7 @@
 // RUN: %clang -S -### -cl-std=CL1.1 %s 2>&1 | FileCheck --check-prefix=CHECK-CL11 %s
 // RUN: %clang -S -### -cl-std=CL1.2 %s 2>&1 | FileCheck --check-prefix=CHECK-CL12 %s
 // RUN: %clang -S -### -cl-std=CL2.0 %s 2>&1 | FileCheck --check-prefix=CHECK-CL20 %s
+// RUN: %clang -S -### -cl-std=CL3.0 %s 2>&1 | FileCheck --check-prefix=CHECK-CL30 %s
 // RUN: %clang -S -### -cl-std=clc++ %s 2>&1 | FileCheck --check-prefix=CHECK-CLCPP %s
 // RUN: %clang -S -### -cl-opt-disable %s 2>&1 | FileCheck --check-prefix=CHECK-OPT-DISABLE %s
 // RUN: %clang -S -### -cl-strict-aliasing %s 2>&1 | FileCheck --check-prefix=CHECK-STRICT-ALIASING %s
@@ -22,6 +23,7 @@
 // CHECK-CL11: "-cc1" {{.*}} "-cl-std=CL1.1"
 // CHECK-CL12: "-cc1" {{.*}} "-cl-std=CL1.2"
 // CHECK-CL20: "-cc1" {{.*}} "-cl-std=CL2.0"
+// CHECK-CL30: "-cc1" {{.*}} "-cl-std=CL3.0"
 // CHECK-CLCPP: "-cc1" {{.*}} "-cl-std=clc++"
 // CHECK-OPT-DISABLE: "-cc1" {{.*}} "-cl-opt-disable"
 // CHECK-STRICT-ALIASING: "-cc1" {{.*}} "-cl-strict-aliasing"
diff --git a/clang/test/Driver/unknown-std.cl b/clang/test/Driver/unknown-std.cl
index 6f371bac13ac..00209fb62556 100644
--- a/clang/test/Driver/unknown-std.cl
+++ b/clang/test/Driver/unknown-std.cl
@@ -10,6 +10,7 @@
 // CHECK-NEXT: note: use 'cl1.1' for 'OpenCL 1.1' standard
 // CHECK-NEXT: note: use 'cl1.2' for 'OpenCL 1.2' standard
 // CHECK-NEXT: note: use 'cl2.0' for 'OpenCL 2.0' standard
+// CHECK-NEXT: note: use 'cl3.0' for 'OpenCL 3.0' standard
 // CHECK-NEXT: note: use 'clc++' for 'C++ for OpenCL' standard
 
 // Make sure that no other output is present.
diff --git a/clang/test/Frontend/stdlang.c b/clang/test/Frontend/stdlang.c
index 51484999e37a..eac4632fbdd6 100644
--- a/clang/test/Frontend/stdlang.c
+++ b/clang/test/Frontend/stdlang.c
@@ -9,6 +9,7 @@
 // RUN: %clang_cc1 -x cl -cl-std=CL1.1 -DOPENCL %s
 // RUN: %clang_cc1 -x cl -cl-std=CL1.2 -DOPENCL %s
 // RUN: %clang_cc1 -x cl -cl-std=CL2.0 -DOPENCL %s
+// RUN: %clang_cc1 -x cl -cl-std=CL3.0 -DOPENCL %s
 // RUN: %clang_cc1 -x cl -cl-std=CLC++ -DOPENCL %s
 // RUN: not %clang_cc1 -x cl -std=c99 -DOPENCL %s 2>&1 | FileCheck --check-prefix=CHECK-C99 %s
 // RUN: not %clang_cc1 -x cl -cl-std=invalid -DOPENCL %s 2>&1 | FileCheck --check-prefix=CHECK-INVALID %s
diff --git a/clang/test/Headers/opencl-c-header.cl b/clang/test/Headers/opencl-c-header.cl
index 1b151ffdd16a..2716076acdcf 100644
--- a/clang/test/Headers/opencl-c-header.cl
+++ b/clang/test/Headers/opencl-c-header.cl
@@ -1,6 +1,7 @@
 // RUN: %clang_cc1 -O0 -triple spir-unknown-unknown -internal-isystem ../../lib/Headers -include opencl-c.h -emit-llvm -o - %s -verify | FileCheck %s
 // RUN: %clang_cc1 -O0 -triple spir-unknown-unknown -internal-isystem ../../lib/Headers -include opencl-c.h -emit-llvm -o - %s -verify -cl-std=CL1.1 | FileCheck %s
 // RUN: %clang_cc1 -O0 -triple spir-unknown-unknown -internal-isystem ../../lib/Headers -include opencl-c.h -emit-llvm -o - %s -verify -cl-std=CL1.2 | FileCheck %s
+// RUN: %clang_cc1 -O0 -triple spir-unknown-unknown -internal-isystem ../../lib/Headers -include opencl-c.h -emit-llvm -o - %s -verify -cl-std=CL3.0 | FileCheck %s
 // RUN: %clang_cc1 -O0 -triple spir-unknown-unknown -internal-isystem ../../lib/Headers -include opencl-c.h -emit-llvm -o - %s -verify -cl-std=clc++ | FileCheck %s --check-prefix=CHECK20
 
 // Test including the default header as a module.
@@ -39,9 +40,11 @@
 // RUN: rm -rf %t
 // RUN: mkdir -p %t
 // RUN: %clang_cc1 -triple spir64-unknown-unknown -emit-llvm -o - -cl-std=CL1.2 -finclude-default-header -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -ftime-report %s 2>&1 | FileCheck --check-prefix=CHECK --check-prefix=CHECK-MOD %s
+// RUN: %clang_cc1 -triple spir64-unknown-unknown -emit-llvm -o - -cl-std=CL3.0 -finclude-default-header -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -ftime-report %s 2>&1 | FileCheck --check-prefix=CHECK --check-prefix=CHECK-MOD %s
 // RUN: %clang_cc1 -triple amdgcn--amdhsa -O0 -emit-llvm -o - -cl-std=CL2.0 -finclude-default-header -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -ftime-report %s 2>&1 | FileCheck --check-prefix=CHECK20 --check-prefix=CHECK-MOD %s
 // RUN: chmod u-w %t 
 // RUN: %clang_cc1 -triple spir64-unknown-unknown -emit-llvm -o - -cl-std=CL1.2 -finclude-default-header -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -ftime-report %s 2>&1 | FileCheck --check-prefix=CHECK --check-prefix=CHECK-MOD %s
+// RUN: %clang_cc1 -triple spir64-unknown-unknown -emit-llvm -o - -cl-std=CL3.0 -finclude-default-header -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -ftime-report %s 2>&1 | FileCheck --check-prefix=CHECK --check-prefix=CHECK-MOD %s
 // RUN: %clang_cc1 -triple amdgcn--amdhsa -O0 -emit-llvm -o - -cl-std=CL2.0 -finclude-default-header -fmodules -fimplicit-module-maps -fmodules-cache-path=%t -ftime-report %s 2>&1 | FileCheck --check-prefix=CHECK20 --check-prefix=CHECK-MOD %s
 // RUN: chmod u+w %t
 
@@ -67,7 +70,7 @@ char f(char x) {
 // from OpenCL 2.0 onwards.
 
 // CHECK20: _Z12write_imagef14ocl_image3d_wo
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ == CL_VERSION_2_0)
 void test_image3dwo(write_only image3d_t img) {
   write_imagef(img, (0), (0.0f));
 }
@@ -81,7 +84,7 @@ void test_atomics(__generic volatile unsigned int* a) {
 #endif
 
 // Verify that ATOMIC_VAR_INIT is defined.
-#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ == CL_VERSION_2_0)
 global atomic_int z = ATOMIC_VAR_INIT(99);
 #endif //__OPENCL_C_VERSION__
 
diff --git a/clang/test/Index/pipe-size.cl b/clang/test/Index/pipe-size.cl
index 94a1255f0a48..59b76051eda1 100644
--- a/clang/test/Index/pipe-size.cl
+++ b/clang/test/Index/pipe-size.cl
@@ -2,6 +2,13 @@
 // RUN: %clang_cc1 -x cl -O0 -cl-std=CL2.0 -emit-llvm -triple spir-unknown-unknown %s -o - | FileCheck %s --check-prefix=SPIR
 // RUN: %clang_cc1 -x cl -O0 -cl-std=CL2.0 -emit-llvm -triple spir64-unknown-unknown %s -o - | FileCheck %s --check-prefix=SPIR64
 // RUN: %clang_cc1 -x cl -O0 -cl-std=CL2.0 -emit-llvm -triple amdgcn-amd-amdhsa %s -o - | FileCheck %s --check-prefix=AMDGCN
+// RUN: %clang_cc1 -x cl -O0 -cl-std=CL3.0 -cl-ext=__opencl_c_pipes -emit-llvm -triple x86_64-unknown-linux-gnu %s -o - | FileCheck %s --check-prefix=X86
+// RUN: %clang_cc1 -x cl -O0 -cl-std=CL3.0 -cl-ext=__opencl_c_pipes -emit-llvm -triple spir-unknown-unknown %s -o - | FileCheck %s --check-prefix=SPIR
+// RUN: %clang_cc1 -x cl -O0 -cl-std=CL3.0 -cl-ext=__opencl_c_pipes -emit-llvm -triple spir64-unknown-unknown %s -o - | FileCheck %s --check-prefix=SPIR64
+// RUN: %clang_cc1 -x cl -O0 -cl-std=CL3.0 -cl-ext=__opencl_c_pipes -emit-llvm -triple amdgcn-amd-amdhsa %s -o - | FileCheck %s --check-prefix=AMDGCN
+
+
+
 __kernel void testPipe( pipe int test )
 {
     int s = sizeof(test);
diff --git a/clang/test/Preprocessor/predefined-macros.c b/clang/test/Preprocessor/predefined-macros.c
index def105f4c52e..b088a37ba665 100644
--- a/clang/test/Preprocessor/predefined-macros.c
+++ b/clang/test/Preprocessor/predefined-macros.c
@@ -129,6 +129,8 @@
 // RUN:   | FileCheck -match-full-lines %s --check-prefix=CHECK-CL12
 // RUN: %clang_cc1 %s -E -dM -o - -x cl -cl-std=CL2.0 \
 // RUN:   | FileCheck -match-full-lines %s --check-prefix=CHECK-CL20
+// RUN: %clang_cc1 %s -E -dM -o - -x cl -cl-std=CL3.0 \
+// RUN:   | FileCheck -match-full-lines %s --check-prefix=CHECK-CL30
 // RUN: %clang_cc1 %s -E -dM -o - -x cl -cl-fast-relaxed-math \
 // RUN:   | FileCheck -match-full-lines %s --check-prefix=CHECK-FRM
 // RUN: %clang_cc1 %s -E -dM -o - -x cl -cl-std=clc++ \
@@ -137,26 +139,37 @@
 // CHECK-CL10: #define CL_VERSION_1_1 110
 // CHECK-CL10: #define CL_VERSION_1_2 120
 // CHECK-CL10: #define CL_VERSION_2_0 200
+// CHECK-CL10: #define CL_VERSION_3_0 300
 // CHECK-CL10: #define __OPENCL_C_VERSION__ 100
 // CHECK-CL10-NOT: #define __FAST_RELAXED_MATH__ 1
 // CHECK-CL11: #define CL_VERSION_1_0 100
 // CHECK-CL11: #define CL_VERSION_1_1 110
 // CHECK-CL11: #define CL_VERSION_1_2 120
 // CHECK-CL11: #define CL_VERSION_2_0 200
+// CHECK-CL11: #define CL_VERSION_3_0 300
 // CHECK-CL11: #define __OPENCL_C_VERSION__ 110
 // CHECK-CL11-NOT: #define __FAST_RELAXED_MATH__ 1
 // CHECK-CL12: #define CL_VERSION_1_0 100
 // CHECK-CL12: #define CL_VERSION_1_1 110
 // CHECK-CL12: #define CL_VERSION_1_2 120
 // CHECK-CL12: #define CL_VERSION_2_0 200
+// CHECK-CL12: #define CL_VERSION_3_0 300
 // CHECK-CL12: #define __OPENCL_C_VERSION__ 120
 // CHECK-CL12-NOT: #define __FAST_RELAXED_MATH__ 1
 // CHECK-CL20: #define CL_VERSION_1_0 100
 // CHECK-CL20: #define CL_VERSION_1_1 110
 // CHECK-CL20: #define CL_VERSION_1_2 120
 // CHECK-CL20: #define CL_VERSION_2_0 200
+// CHECK-CL20: #define CL_VERSION_3_0 300
 // CHECK-CL20: #define __OPENCL_C_VERSION__ 200
 // CHECK-CL20-NOT: #define __FAST_RELAXED_MATH__ 1
+// CHECK-CL30: #define CL_VERSION_1_0 100
+// CHECK-CL30: #define CL_VERSION_1_1 110
+// CHECK-CL30: #define CL_VERSION_1_2 120
+// CHECK-CL30: #define CL_VERSION_2_0 200
+// CHECK-CL30: #define CL_VERSION_3_0 300
+// CHECK-CL30: #define __OPENCL_C_VERSION__ 300
+// CHECK-CL30-NOT: #define __FAST_RELAXED_MATH__ 1
 // CHECK-FRM: #define __FAST_RELAXED_MATH__ 1
 // CHECK-CLCPP10: #define __CL_CPP_VERSION_1_0__ 100
 // CHECK-CLCPP10: #define __OPENCL_CPP_VERSION__ 100
diff --git a/clang/test/Sema/feature-extensions-simult-support.cl b/clang/test/Sema/feature-extensions-simult-support.cl
new file mode 100644
index 000000000000..0789105002b2
--- /dev/null
+++ b/clang/test/Sema/feature-extensions-simult-support.cl
@@ -0,0 +1,75 @@
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL1.2 -cl-ext=-cl_khr_fp64
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL1.2 -cl-ext=+cl_khr_fp64
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL1.2 -cl-ext=-cl_khr_3d_image_writes
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL1.2 -cl-ext=+cl_khr_3d_image_writes
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL1.2 -cl-ext=-cl_khr_subgroups
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL1.2 -cl-ext=+cl_khr_subgroups
+
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -cl-ext=-cl_khr_fp64
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -cl-ext=+cl_khr_fp64
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -cl-ext=-cl_khr_3d_image_writes
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -cl-ext=+cl_khr_3d_image_writes
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -cl-ext=-cl_khr_subgroups
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -cl-ext=+cl_khr_subgroups
+
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_fp64
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_3d_image_writes
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_subgroups
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_pipes
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_device_enqueue
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_read_write_images
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_fp64,-cl_khr_fp64
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_3d_image_writes,-cl_khr_3d_image_writes
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+__opencl_c_subgroups,-cl_khr_subgroups
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+cl_khr_fp64
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+cl_khr_3d_image_writes
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=+cl_khr_subgroups
+
+// expected-no-diagnostics
+
+#ifdef cl_khr_fp64
+  #ifndef __opencl_c_fp64
+    #error macros were not properly set up
+  #endif
+#endif
+#ifdef __opencl_c_fp64
+  #ifndef cl_khr_fp64
+    #error macros were not properly set up
+  #endif
+#endif
+
+#ifdef cl_khr_3d_image_writes
+  #ifndef __opencl_c_3d_image_writes
+    #error macros were not properly set up
+  #endif
+#endif
+#ifdef __opencl_c_3d_image_writes
+  #ifndef cl_khr_3d_image_writes
+    #error macros were not properly set up
+  #endif
+#endif
+
+#ifdef cl_khr_subgroups
+  #ifndef __opencl_c_subgroups
+    #error macros were not properly set up
+  #endif
+#endif
+#ifdef __opencl_c_subgroups
+  #ifndef cl_khr_subgroups
+    #error macros were not properly set up
+  #endif
+#endif
+
+#if defined(__opencl_c_pipes) || defined(__opencl_c_device_enqueue)
+  #ifndef __opencl_c_generic_address_space
+    #error macros were not properly set up
+  #endif
+#endif
+
+#if defined(__opencl_c_3d_image_writes) || defined(__opencl_c_read_write_images)
+  #ifndef __opencl_c_images
+    #error macros were not properly set up
+  #endif
+#endif
+
+kernel void test(){}
diff --git a/clang/test/Sema/features-ignore-pragma.cl b/clang/test/Sema/features-ignore-pragma.cl
new file mode 100644
index 000000000000..046ce5390754
--- /dev/null
+++ b/clang/test/Sema/features-ignore-pragma.cl
@@ -0,0 +1,24 @@
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=__opencl_c_fp64
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=__opencl_c_subgroups
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=__opencl_c_3d_image_writes
+
+#pragma OPENCL EXTENSION __opencl_c_fp64 : enable
+// expected-warning@-1 {{OpenCL feature support can't be controlled via pragma, ignoring}}
+
+#pragma OPENCL EXTENSION cl_khr_fp64 : enable
+#ifndef __opencl_c_fp64
+// expected-warning@-2{{unsupported OpenCL extension 'cl_khr_fp64' - ignoring}}
+#endif
+
+#pragma OPENCL EXTENSION cl_khr_subgroups : enable
+#ifndef __opencl_c_subgroups
+// expected-warning@-2{{unsupported OpenCL extension 'cl_khr_subgroups' - ignoring}}
+#endif
+
+#pragma OPENCL EXTENSION cl_khr_3d_image_writes : enable
+#ifndef __opencl_c_3d_image_writes
+// expected-warning@-2{{unsupported OpenCL extension 'cl_khr_3d_image_writes' - ignoring}}
+#endif
+
+kernel void foo() {}
diff --git a/clang/test/Sema/opencl-features-pipes.cl b/clang/test/Sema/opencl-features-pipes.cl
new file mode 100644
index 000000000000..c0ac778f24a6
--- /dev/null
+++ b/clang/test/Sema/opencl-features-pipes.cl
@@ -0,0 +1,18 @@
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL1.1
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL1.2
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -DHAS
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=__opencl_c_pipes -DHAS
+// expected-no-diagnostics
+
+#ifdef HAS
+  #ifndef __opencl_c_pipes
+   #error Feature should be defined
+  #endif
+#else
+  #ifdef __opencl_c_pipes
+    #error Feature should not be defined
+  #endif
+#endif
+
+kernel void foo() {}
diff --git a/clang/test/Sema/opencl-features.cl b/clang/test/Sema/opencl-features.cl
new file mode 100644
index 000000000000..aa432f6b60bf
--- /dev/null
+++ b/clang/test/Sema/opencl-features.cl
@@ -0,0 +1,128 @@
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -cl-ext=-__opencl_c_device_enqueue,-__opencl_c_pipes,-__opencl_c_read_write_images
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CLC++
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0
+// expected-no-diagnostics
+
+#ifndef __opencl_c_int64
+  #error Feature __opencl_c_int64 shouldn't be defined
+#endif
+
+#if defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ == CL_VERSION_2_0)
+  #ifndef __opencl_c_3d_image_writes
+   #error Feature __opencl_c_3d_image_writes should be defined
+  #endif
+
+  #ifndef __opencl_c_atomic_order_acq_rel
+   #error Feature __opencl_c_atomic_order_acq_rel should be defined
+  #endif
+
+  #ifndef __opencl_c_atomic_order_seq_cst
+   #error Feature __opencl_c_atomic_order_seq_cst should be defined
+  #endif
+
+  #ifndef __opencl_c_atomic_scope_device
+   #error Feature __opencl_c_atomic_scope_device should be defined
+  #endif
+
+  #ifndef __opencl_c_atomic_scope_all_devices
+   #error Feature __opencl_c_atomic_scope_all_devices should be defined
+  #endif
+
+  #ifndef __opencl_c_device_enqueue
+     #error Feature __opencl_c_device_enqueue should be defined
+  #endif
+
+  #ifndef __opencl_c_generic_address_space
+   #error Feature __opencl_c_generic_address_space should be defined
+  #endif
+
+  #ifndef __opencl_c_pipes
+   #error Feature __opencl_c_pipes should be defined
+  #endif
+
+  #ifndef __opencl_c_program_scope_global_variables
+   #error Feature __opencl_c_program_scope_global_variables should be defined
+  #endif
+
+  #ifndef __opencl_c_read_write_images
+   #error Feature __opencl_c_read_write_images should be defined
+  #endif
+
+  #ifndef __opencl_c_subgroups
+   #error Feature __opencl_c_subgroups should be defined
+  #endif
+
+  #ifndef __opencl_c_work_group_collective_functions
+   #error Feature __opencl_c_work_group_collective_functions should be defined
+  #endif
+
+  #ifndef __opencl_c_fp64
+   #error Feature __opencl_c_fp64 should be defined
+  #endif
+
+  #ifndef __opencl_c_images
+   #error Feature __opencl_c_images should be defined
+  #endif
+#endif // defined(__OPENCL_CPP_VERSION__) || (__OPENCL_C_VERSION__ >= CL_VERSION_2_0)
+
+
+#if __OPENCL_C_VERSION__ == CL_VERSION_3_0
+  #ifdef __opencl_c_3d_image_writes
+   #error Feature __opencl_c_3d_image_writes shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_atomic_order_acq_rel
+   #error Feature __opencl_c_atomic_order_acq_rel shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_atomic_order_seq_cst
+   #error Feature __opencl_c_atomic_order_seq_cst shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_atomic_scope_device
+   #error Feature __opencl_c_atomic_scope_device shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_atomic_scope_all_devices
+   #error Feature __opencl_c_atomic_scope_all_devices shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_device_enqueue
+   #error Feature __opencl_c_device_enqueue shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_generic_address_space
+   #error Feature __opencl_c_generic_address_space shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_pipes
+   #error Feature __opencl_c_pipes shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_program_scope_global_variables
+   #error Feature __opencl_c_program_scope_global_variables shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_read_write_images
+   #error Feature __opencl_c_read_write_images shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_subgroups
+   #error Feature __opencl_c_subgroups shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_work_group_collective_functions
+   #error Feature __opencl_c_work_group_collective_functions shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_fp64
+   #error Feature __opencl_c_fp64 shouldn't be defined
+  #endif
+
+  #ifdef __opencl_c_images
+   #error Feature __opencl_c_images shouldn't be defined
+  #endif
+#endif // __OPENCL_C_VERSION__ == CL_VERSION_3_0
+
+kernel void foo() {}
diff --git a/clang/test/Sema/pipe_builtins_feature.cl b/clang/test/Sema/pipe_builtins_feature.cl
new file mode 100644
index 000000000000..56fa94fc7705
--- /dev/null
+++ b/clang/test/Sema/pipe_builtins_feature.cl
@@ -0,0 +1,21 @@
+// RUN: %clang_cc1 -cl-std=CL2.0 -fsyntax-only -verify %s
+// RUN: %clang_cc1 -cl-std=CL3.0 -fsyntax-only -verify %s
+// RUN: %clang_cc1 -cl-std=CL3.0 -cl-ext=__opencl_c_pipes -fsyntax-only -verify %s
+
+#ifdef __opencl_c_pipes
+  #ifndef __opencl_c_generic_address_space
+    #error Generic address space feature must also be defined
+  #endif
+// CHECK: expected-no-diagnostics
+// check that pragma disable all doesn't touch feature support
+  #pragma OPENCL EXTENSION all : disable
+#endif
+
+void test(read_only pipe int p, global int *ptr) {
+    reserve_id_t rid;
+}
+
+#ifndef __opencl_c_pipes
+// expected-error@-5 {{expected parameter declarator}} expected-error@-5 {{expected ')'}} expected-note@-5 {{to match this '('}}
+// expected-error@-5 {{use of type 'reserve_id_t' requires __opencl_c_pipes extension to be enabled}}
+#endif
diff --git a/clang/test/SemaOpenCL/address-spaces-conversions-cl2.0.cl b/clang/test/SemaOpenCL/address-spaces-conversions-cl2.0.cl
index 5efea216346a..09aea27ae6de 100644
--- a/clang/test/SemaOpenCL/address-spaces-conversions-cl2.0.cl
+++ b/clang/test/SemaOpenCL/address-spaces-conversions-cl2.0.cl
@@ -1,6 +1,9 @@
 // RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DCONSTANT -cl-std=CL2.0
 // RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DGLOBAL -cl-std=CL2.0
 // RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DGENERIC -cl-std=CL2.0
+// RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DCONSTANT -cl-std=CL3.0 -cl-ext=__opencl_c_generic_address_space
+// RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DGLOBAL -cl-std=CL3.0 -cl-ext=__opencl_c_generic_address_space
+// RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DGENERIC -cl-std=CL3.0 -cl-ext=__opencl_c_generic_address_space
 // RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DCONSTANT -cl-std=clc++
 // RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DGLOBAL -cl-std=clc++
 // RUN: %clang_cc1 %s -ffake-address-space-map -verify -pedantic -fsyntax-only -DGENERIC -cl-std=clc++
diff --git a/clang/test/SemaOpenCL/address-spaces.cl b/clang/test/SemaOpenCL/address-spaces.cl
index 07547ea19680..e367a7c57292 100644
--- a/clang/test/SemaOpenCL/address-spaces.cl
+++ b/clang/test/SemaOpenCL/address-spaces.cl
@@ -1,5 +1,6 @@
 // RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only
 // RUN: %clang_cc1 %s -cl-std=CL2.0 -verify -pedantic -fsyntax-only
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -cl-ext=__opencl_c_generic_address_space -verify -pedantic -fsyntax-only
 // RUN: %clang_cc1 %s -cl-std=clc++ -verify -pedantic -fsyntax-only
 
 __constant int ci = 1;
diff --git a/clang/test/SemaOpenCL/cl20-device-side-enqueue.cl b/clang/test/SemaOpenCL/cl20-device-side-enqueue.cl
index f63e2913c749..727141190a0b 100644
--- a/clang/test/SemaOpenCL/cl20-device-side-enqueue.cl
+++ b/clang/test/SemaOpenCL/cl20-device-side-enqueue.cl
@@ -2,6 +2,12 @@
 // RUN: %clang_cc1 %s -cl-std=CL2.0 -triple "spir-unknown-unknown" -verify -pedantic -fsyntax-only -DB32 -DQUALS="const volatile"
 // RUN: %clang_cc1 %s -cl-std=CL2.0 -triple "spir64-unknown-unknown" -verify -pedantic -fsyntax-only -Wconversion -DWCONV -DQUALS=
 // RUN: %clang_cc1 %s -cl-std=CL2.0 -triple "spir64-unknown-unknown" -verify -pedantic -fsyntax-only -Wconversion -DWCONV -DQUALS="const volatile"
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir-unknown-unknown" -cl-ext=__opencl_c_device_enqueue,__opencl_c_subgroups  -verify -pedantic -fsyntax-only -DB32 -DQUALS=
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir-unknown-unknown" -cl-ext=__opencl_c_device_enqueue,__opencl_c_subgroups -verify -pedantic -fsyntax-only -DB32 -DQUALS="const volatile"
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir64-unknown-unknown" -cl-ext=__opencl_c_device_enqueue,__opencl_c_subgroups -verify -pedantic -fsyntax-only -Wconversion -DWCONV -DQUALS=
+// RUN: %clang_cc1 %s -cl-std=CL3.0 -triple "spir64-unknown-unknown" -cl-ext=__opencl_c_device_enqueue,__opencl_c_subgroups -verify -pedantic -fsyntax-only -Wconversion -DWCONV -DQUALS="const volatile"
+
+
 
 typedef struct {int a;} ndrange_t;
 // Diagnostic tests for different overloads of enqueue_kernel from Table 6.13.17.1 of OpenCL 2.0 Spec.
@@ -235,11 +241,17 @@ kernel void bar(global unsigned int *buf)
 kernel void foo1(global unsigned int *buf)
 {
   ndrange_t n;
-  buf[0] = get_kernel_max_sub_group_size_for_ndrange(n, ^(){}); // expected-error {{use of declaration 'get_kernel_max_sub_group_size_for_ndrange' requires cl_khr_subgroups extension to be enabled}}
+  buf[0] = get_kernel_max_sub_group_size_for_ndrange(n, ^(){});
+#if __OPENCL_C_VERSION__ < 300
+// expected-error@-2 {{use of declaration 'get_kernel_max_sub_group_size_for_ndrange' requires cl_khr_subgroups extension to be enabled}}
+#endif
 }
 
 kernel void bar1(global unsigned int *buf)
 {
   ndrange_t n;
-  buf[0] = get_kernel_sub_group_count_for_ndrange(n, ^(){}); // expected-error {{use of declaration 'get_kernel_sub_group_count_for_ndrange' requires cl_khr_subgroups extension to be enabled}}
+  buf[0] = get_kernel_sub_group_count_for_ndrange(n, ^(){});
+#if __OPENCL_C_VERSION__ < 300
+// expected-error@-2 {{use of declaration 'get_kernel_sub_group_count_for_ndrange' requires cl_khr_subgroups extension to be enabled}}
+#endif
 }
diff --git a/clang/test/SemaOpenCL/forget-unsupported-builtins.cl b/clang/test/SemaOpenCL/forget-unsupported-builtins.cl
new file mode 100644
index 000000000000..14dd03e2c7db
--- /dev/null
+++ b/clang/test/SemaOpenCL/forget-unsupported-builtins.cl
@@ -0,0 +1,22 @@
+// RUN: %clang_cc1 -cl-std=cl3.0 -fsyntax-only  -verify %s -triple spir-unknown-unknown
+// RUN: %clang_cc1 -cl-std=cl3.0 -fsyntax-only -cl-ext=__opencl_c_pipes,__opencl_c_generic_address_space,__opencl_c_device_enqueue -verify %s -triple spir-unknown-unknown -DFEATURES
+
+#ifndef FEATURES
+ // expected-no-diagnostics
+#else
+ // expected-error@+10 {{cannot redeclare builtin function 'get_pipe_max_packets'}}
+ // expected-note@+9 {{'get_pipe_max_packets' is a builtin with type 'unsigned int ()'}}
+ // expected-error@+9 {{cannot redeclare builtin function 'to_local'}}
+ // expected-note@+8 {{'to_local' is a builtin with type 'void *(void *)'}}
+ // expected-error@+8  {{cannot redeclare builtin function 'to_global'}}
+ // expected-note@+7 {{'to_global' is a builtin with type 'void *(void *)'}}
+ // expected-error@+7 {{cannot redeclare builtin function 'get_kernel_work_group_size'}}
+ // expected-note@+6 {{'get_kernel_work_group_size' is a builtin with type 'unsigned int ()'}}
+#endif
+
+int get_pipe_max_packets(int);
+int to_local(int);
+int to_global(int);
+int get_kernel_work_group_size(int);
+
+kernel void test(global int *dst) {}
diff --git a/clang/test/SemaOpenCL/image-features.cl b/clang/test/SemaOpenCL/image-features.cl
new file mode 100644
index 000000000000..ace6913bb31e
--- /dev/null
+++ b/clang/test/SemaOpenCL/image-features.cl
@@ -0,0 +1,20 @@
+// RUN: %clang_cc1 -cl-std=cl2.0 -fsyntax-only -verify %s -triple spir-unknown-unknown
+// RUN: %clang_cc1 -cl-std=cl3.0 -cl-ext=__opencl_c_images -fsyntax-only -verify %s -triple spir-unknown-unknown
+// RUN: %clang_cc1 -cl-std=cl3.0 -cl-ext=__opencl_c_3d_image_writes -fsyntax-only -verify %s -triple spir-unknown-unknown
+// RUN: %clang_cc1 -cl-std=cl3.0 -cl-ext=__opencl_c_read_write_images -fsyntax-only -verify %s -triple spir-unknown-unknown
+
+#if defined(__opencl_c_read_write_images) && defined(__opencl_c_3d_image_writes)
+ // expected-no-diagnostics
+#endif
+
+__kernel void write_3d_image(__write_only image3d_t i) {}
+
+#ifndef __opencl_c_3d_image_writes
+  // expected-error@-3 {{use of type '__write_only image3d_t' requires __opencl_c_3d_image_writes extension to be enabled}}
+#endif
+
+__kernel void read_write_3d_image(__read_write image3d_t i) { }
+
+#ifndef __opencl_c_read_write_images
+  // expected-error@-3 {{use of type '__read_write image3d_t' requires __opencl_c_read_write_images extension to be enabled}}
+#endif
diff --git a/clang/test/SemaOpenCL/invalid-pipe-builtin-cl2.0.cl b/clang/test/SemaOpenCL/invalid-pipe-builtin-cl2.0.cl
index 36e76621d24a..38b0a04726e3 100644
--- a/clang/test/SemaOpenCL/invalid-pipe-builtin-cl2.0.cl
+++ b/clang/test/SemaOpenCL/invalid-pipe-builtin-cl2.0.cl
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0 -cl-ext=+cl_khr_subgroups
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=__opencl_c_subgroups,__opencl_c_pipes
 
 #pragma OPENCL EXTENSION cl_khr_subgroups : enable
 
diff --git a/clang/test/SemaOpenCL/storageclass-cl20.cl b/clang/test/SemaOpenCL/storageclass-cl20.cl
index 581701d2a6a5..469c526ebc30 100644
--- a/clang/test/SemaOpenCL/storageclass-cl20.cl
+++ b/clang/test/SemaOpenCL/storageclass-cl20.cl
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL2.0
+// RUN: %clang_cc1 %s -verify -pedantic -fsyntax-only -cl-std=CL3.0 -cl-ext=__opencl_c_program_scope_global_variables,__opencl_c_generic_address_space
 
 int G2 = 0;
 global int G3 = 0;
diff --git a/clang/utils/TableGen/ClangOpenCLBuiltinEmitter.cpp b/clang/utils/TableGen/ClangOpenCLBuiltinEmitter.cpp
index 41d33b550680..f50652493e5e 100644
--- a/clang/utils/TableGen/ClangOpenCLBuiltinEmitter.cpp
+++ b/clang/utils/TableGen/ClangOpenCLBuiltinEmitter.cpp
@@ -56,6 +56,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "TableGenBackends.h"
+#include "clang/Basic/OpenCLOptions.h"
 #include "llvm/ADT/MapVector.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallString.h"
@@ -69,6 +70,7 @@
 #include "llvm/TableGen/Record.h"
 #include "llvm/TableGen/StringMatcher.h"
 #include "llvm/TableGen/TableGenBackend.h"
+#include <numeric>
 #include <set>
 
 using namespace llvm;
@@ -228,6 +230,10 @@ private:
   // The function "tan", having the same signatures, would be mapped to the
   // same entry (<I1, I2, I3>).
   MapVector<BuiltinIndexListTy *, BuiltinTableEntries> SignatureListMap;
+
+  // Encode all versions
+  unsigned short
+  EncodeBuiltinVersions(std::vector<Record *> BuiltinVersionsRecords) const;
 };
 } // namespace
 
@@ -338,12 +344,10 @@ struct OpenCLBuiltinStruct {
   const bool IsConst;
   // Function attribute __attribute__((convergent))
   const bool IsConv;
+  // All opencl versions encoded
+  const unsigned char AllVersions : 5;
   // OpenCL extension(s) required for this overload.
   const unsigned short Extension;
-  // First OpenCL version in which this overload was introduced (e.g. CL20).
-  const unsigned short MinVersion;
-  // First OpenCL version in which this overload was removed (e.g. CL20).
-  const unsigned short MaxVersion;
 };
 
 )";
@@ -503,11 +507,9 @@ void BuiltinNameEmitter::EmitBuiltinTable() {
          << (Overload.first->getValueAsBit("IsPure")) << ", "
          << (Overload.first->getValueAsBit("IsConst")) << ", "
          << (Overload.first->getValueAsBit("IsConv")) << ", "
-         << FunctionExtensionIndex[ExtName] << ", "
-         << Overload.first->getValueAsDef("MinVersion")->getValueAsInt("ID")
-         << ", "
-         << Overload.first->getValueAsDef("MaxVersion")->getValueAsInt("ID")
-         << " },\n";
+         << EncodeBuiltinVersions(
+                Overload.first->getValueAsListOfDefs("Versions"))
+         << ", " << FunctionExtensionIndex[ExtName] << " },\n";
       Index++;
     }
   }
@@ -528,10 +530,8 @@ bool BuiltinNameEmitter::CanReuseSignature(
     if (Rec->getValueAsBit("IsPure") == Rec2->getValueAsBit("IsPure") &&
         Rec->getValueAsBit("IsConst") == Rec2->getValueAsBit("IsConst") &&
         Rec->getValueAsBit("IsConv") == Rec2->getValueAsBit("IsConv") &&
-        Rec->getValueAsDef("MinVersion")->getValueAsInt("ID") ==
-            Rec2->getValueAsDef("MinVersion")->getValueAsInt("ID") &&
-        Rec->getValueAsDef("MaxVersion")->getValueAsInt("ID") ==
-            Rec2->getValueAsDef("MaxVersion")->getValueAsInt("ID") &&
+        EncodeBuiltinVersions(Rec->getValueAsListOfDefs("Versions")) ==
+            EncodeBuiltinVersions(Rec2->getValueAsListOfDefs("Versions")) &&
         Rec->getValueAsDef("Extension")->getName() ==
             Rec2->getValueAsDef("Extension")->getName()) {
       return true;
@@ -806,6 +806,15 @@ static void OCL2Qual(ASTContext &Context, const OpenCLTypeStruct &Ty,
   OS << "\n} // OCL2Qual\n";
 }
 
+unsigned short BuiltinNameEmitter::EncodeBuiltinVersions(
+    std::vector<Record *> BuiltinVersionsRecords) const {
+  return std::accumulate(
+      BuiltinVersionsRecords.begin(), BuiltinVersionsRecords.end(),
+      (unsigned short)0, [](unsigned short C, Record *R) {
+        return C |= clang::OpenCLOptions::EncodeVersion(R->getValueAsInt("ID"));
+      });
+}
+
 void clang::EmitClangOpenCLBuiltins(RecordKeeper &Records, raw_ostream &OS) {
   BuiltinNameEmitter NameChecker(Records, OS);
   NameChecker.Emit();
-- 
2.17.1