arsenm updated this revision to Diff 257684.
arsenm added a comment.
Attach correct diff
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D77923/new/
https://reviews.llvm.org/D77923
Files:
clang/include/clang/Basic/TargetInfo.h
clang/lib/Basic/Targets/AMDGPU.cpp
clang/lib/Basic/Targets/AMDGPU.h
clang/lib/Basic/Targets/SPIR.h
clang/lib/Frontend/InitPreprocessor.cpp
clang/test/Preprocessor/predefined-macros.c
llvm/include/llvm/Support/TargetParser.h
llvm/lib/Support/TargetParser.cpp
Index: llvm/lib/Support/TargetParser.cpp
===================================================================
--- llvm/lib/Support/TargetParser.cpp
+++ llvm/lib/Support/TargetParser.cpp
@@ -72,36 +72,36 @@
{{"oland"}, {"gfx601"}, GK_GFX601, FEATURE_NONE},
{{"pitcairn"}, {"gfx601"}, GK_GFX601, FEATURE_NONE},
{{"verde"}, {"gfx601"}, GK_GFX601, FEATURE_NONE},
- {{"gfx700"}, {"gfx700"}, GK_GFX700, FEATURE_NONE},
- {{"kaveri"}, {"gfx700"}, GK_GFX700, FEATURE_NONE},
- {{"gfx701"}, {"gfx701"}, GK_GFX701, FEATURE_FAST_FMA_F32},
- {{"hawaii"}, {"gfx701"}, GK_GFX701, FEATURE_FAST_FMA_F32},
- {{"gfx702"}, {"gfx702"}, GK_GFX702, FEATURE_FAST_FMA_F32},
- {{"gfx703"}, {"gfx703"}, GK_GFX703, FEATURE_NONE},
- {{"kabini"}, {"gfx703"}, GK_GFX703, FEATURE_NONE},
- {{"mullins"}, {"gfx703"}, GK_GFX703, FEATURE_NONE},
- {{"gfx704"}, {"gfx704"}, GK_GFX704, FEATURE_NONE},
- {{"bonaire"}, {"gfx704"}, GK_GFX704, FEATURE_NONE},
- {{"gfx801"}, {"gfx801"}, GK_GFX801, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32},
- {{"carrizo"}, {"gfx801"}, GK_GFX801, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32},
- {{"gfx802"}, {"gfx802"}, GK_GFX802, FEATURE_FAST_DENORMAL_F32},
- {{"iceland"}, {"gfx802"}, GK_GFX802, FEATURE_FAST_DENORMAL_F32},
- {{"tonga"}, {"gfx802"}, GK_GFX802, FEATURE_FAST_DENORMAL_F32},
- {{"gfx803"}, {"gfx803"}, GK_GFX803, FEATURE_FAST_DENORMAL_F32},
- {{"fiji"}, {"gfx803"}, GK_GFX803, FEATURE_FAST_DENORMAL_F32},
- {{"polaris10"}, {"gfx803"}, GK_GFX803, FEATURE_FAST_DENORMAL_F32},
- {{"polaris11"}, {"gfx803"}, GK_GFX803, FEATURE_FAST_DENORMAL_F32},
- {{"gfx810"}, {"gfx810"}, GK_GFX810, FEATURE_FAST_DENORMAL_F32},
- {{"stoney"}, {"gfx810"}, GK_GFX810, FEATURE_FAST_DENORMAL_F32},
- {{"gfx900"}, {"gfx900"}, GK_GFX900, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32},
- {{"gfx902"}, {"gfx902"}, GK_GFX902, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32},
- {{"gfx904"}, {"gfx904"}, GK_GFX904, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32},
- {{"gfx906"}, {"gfx906"}, GK_GFX906, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32},
- {{"gfx908"}, {"gfx908"}, GK_GFX908, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32},
- {{"gfx909"}, {"gfx909"}, GK_GFX909, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32},
- {{"gfx1010"}, {"gfx1010"}, GK_GFX1010, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
- {{"gfx1011"}, {"gfx1011"}, GK_GFX1011, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
- {{"gfx1012"}, {"gfx1012"}, GK_GFX1012, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
+ {{"gfx700"}, {"gfx700"}, GK_GFX700, FEATURE_FLAT_ADDRESS_SPACE},
+ {{"kaveri"}, {"gfx700"}, GK_GFX700, FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx701"}, {"gfx701"}, GK_GFX701, FEATURE_FAST_FMA_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"hawaii"}, {"gfx701"}, GK_GFX701, FEATURE_FAST_FMA_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx702"}, {"gfx702"}, GK_GFX702, FEATURE_FAST_FMA_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx703"}, {"gfx703"}, GK_GFX703, FEATURE_FLAT_ADDRESS_SPACE},
+ {{"kabini"}, {"gfx703"}, GK_GFX703, FEATURE_FLAT_ADDRESS_SPACE},
+ {{"mullins"}, {"gfx703"}, GK_GFX703, FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx704"}, {"gfx704"}, GK_GFX704, FEATURE_FLAT_ADDRESS_SPACE},
+ {{"bonaire"}, {"gfx704"}, GK_GFX704, FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx801"}, {"gfx801"}, GK_GFX801, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"carrizo"}, {"gfx801"}, GK_GFX801, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx802"}, {"gfx802"}, GK_GFX802, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"iceland"}, {"gfx802"}, GK_GFX802, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"tonga"}, {"gfx802"}, GK_GFX802, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx803"}, {"gfx803"}, GK_GFX803, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"fiji"}, {"gfx803"}, GK_GFX803, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"polaris10"}, {"gfx803"}, GK_GFX803, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"polaris11"}, {"gfx803"}, GK_GFX803, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx810"}, {"gfx810"}, GK_GFX810, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"stoney"}, {"gfx810"}, GK_GFX810, FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx900"}, {"gfx900"}, GK_GFX900, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx902"}, {"gfx902"}, GK_GFX902, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx904"}, {"gfx904"}, GK_GFX904, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx906"}, {"gfx906"}, GK_GFX906, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx908"}, {"gfx908"}, GK_GFX908, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx909"}, {"gfx909"}, GK_GFX909, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx1010"}, {"gfx1010"}, GK_GFX1010, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx1011"}, {"gfx1011"}, GK_GFX1011, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_FLAT_ADDRESS_SPACE},
+ {{"gfx1012"}, {"gfx1012"}, GK_GFX1012, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_FLAT_ADDRESS_SPACE},
};
const GPUInfo *getArchEntry(AMDGPU::GPUKind AK, ArrayRef<GPUInfo> Table) {
Index: llvm/include/llvm/Support/TargetParser.h
===================================================================
--- llvm/include/llvm/Support/TargetParser.h
+++ llvm/include/llvm/Support/TargetParser.h
@@ -154,7 +154,10 @@
FEATURE_FAST_DENORMAL_F32 = 1 << 5,
// Wavefront 32 is available.
- FEATURE_WAVE32 = 1 << 6
+ FEATURE_WAVE32 = 1 << 6,
+
+ // Flat instructions are available.
+ FEATURE_FLAT_ADDRESS_SPACE = 1 << 7,
};
StringRef getArchNameAMDGCN(GPUKind AK);
Index: clang/test/Preprocessor/predefined-macros.c
===================================================================
--- clang/test/Preprocessor/predefined-macros.c
+++ clang/test/Preprocessor/predefined-macros.c
@@ -174,6 +174,31 @@
// RUN: %clang_cc1 %s -E -dM -o - -x cl -triple spir-unknown-unknown \
// RUN: | FileCheck -match-full-lines %s --check-prefix=CHECK-SPIR
// CHECK-SPIR: #define __IMAGE_SUPPORT__ 1
+// CHECK-SPIR: #define __OPENCL_VERSION__ 200{{$}}
+
+// RUN: %clang_cc1 %s -E -dM -o - -x cl -triple amdgcn-unknown-unknown \
+// RUN: | FileCheck -match-full-lines %s --check-prefix=CHECK-AMDGCN-GFX6
+// RUN: %clang_cc1 %s -E -dM -o - -x cl -triple amdgcn-unknown-unknown -target-cpu gfx600 \
+// RUN: | FileCheck -match-full-lines %s --check-prefix=CHECK-AMDGCN-GFX6
+// CHECK-AMDGCN-GFX6: #define __IMAGE_SUPPORT__ 1
+// CHECK-AMDGCN-GFX6: #define __OPENCL_VERSION__ 120{{$}}
+
+// No set OS or mesa3d, assume CL1.2
+// RUN: %clang_cc1 %s -E -dM -o - -x cl -triple amdgcn-unknown-unknown -target-cpu gfx700 \
+// RUN: | FileCheck -match-full-lines %s --check-prefix=CHECK-AMDGCN-GFX7
+// RUN: %clang_cc1 %s -E -dM -o - -x cl -triple amdgcn-unknown-mesa3d -target-cpu gfx700 \
+// RUN: | FileCheck -match-full-lines %s --check-prefix=CHECK-AMDGCN-GFX7
+// CHECK-AMDGCN-GFX7: #define __IMAGE_SUPPORT__ 1
+// CHECK-AMDGCN-GFX7: #define __OPENCL_VERSION__ 120{{$}}
+
+// Assume CL2.0 support for HSA and PAL
+// RUN: %clang_cc1 %s -E -dM -o - -x cl -triple amdgcn-unknown-amdhsa -target-cpu gfx700 \
+// RUN: | FileCheck -match-full-lines %s --check-prefix=CHECK-AMDGCN-AMDHSAPAL
+// RUN: %clang_cc1 %s -E -dM -o - -x cl -triple amdgcn-unknown-amdpal -target-cpu gfx700 \
+// RUN: | FileCheck -match-full-lines %s --check-prefix=CHECK-AMDGCN-AMDHSAPAL
+// CHECK-AMDGCN-AMDHSAPAL: #define __IMAGE_SUPPORT__ 1
+// CHECK-AMDGCN-AMDHSAPAL: #define __OPENCL_VERSION__ 200{{$}}
+
// RUN: %clang_cc1 %s -E -dM -o - -x hip -triple amdgcn-amd-amdhsa \
// RUN: | FileCheck -match-full-lines %s --check-prefix=CHECK-HIP
Index: clang/lib/Frontend/InitPreprocessor.cpp
===================================================================
--- clang/lib/Frontend/InitPreprocessor.cpp
+++ clang/lib/Frontend/InitPreprocessor.cpp
@@ -449,6 +449,12 @@
llvm_unreachable("Unsupported OpenCL version");
}
}
+
+ if (TI.getMaxOpenCLSupportedVersion() != 0) {
+ Builder.defineMacro("__OPENCL_VERSION__",
+ Twine(TI.getMaxOpenCLSupportedVersion()));
+ }
+
Builder.defineMacro("CL_VERSION_1_0", "100");
Builder.defineMacro("CL_VERSION_1_1", "110");
Builder.defineMacro("CL_VERSION_1_2", "120");
@@ -1112,7 +1118,7 @@
Builder.defineMacro(#Ext);
#include "clang/Basic/OpenCLExtensions.def"
- if (TI.getTriple().isSPIR())
+ if (TI.hasOpenCLImageSupport())
Builder.defineMacro("__IMAGE_SUPPORT__");
}
Index: clang/lib/Basic/Targets/SPIR.h
===================================================================
--- clang/lib/Basic/Targets/SPIR.h
+++ clang/lib/Basic/Targets/SPIR.h
@@ -54,6 +54,8 @@
// Define available target features
// These must be defined in sorted order!
NoAsmVariants = true;
+ SupportsOpenCLImages = true;
+ MaxOpenCLSupportedVersion = 200;
}
void getTargetDefines(const LangOptions &Opts,
Index: clang/lib/Basic/Targets/AMDGPU.h
===================================================================
--- clang/lib/Basic/Targets/AMDGPU.h
+++ clang/lib/Basic/Targets/AMDGPU.h
@@ -70,6 +70,10 @@
!!(GPUFeatures & llvm::AMDGPU::FEATURE_LDEXP);
}
+ bool hasFlatAddressSpace() const {
+ return !!(GPUFeatures & llvm::AMDGPU::FEATURE_FLAT_ADDRESS_SPACE);
+ }
+
static bool isAMDGCN(const llvm::Triple &TT) {
return TT.getArch() == llvm::Triple::amdgcn;
}
Index: clang/lib/Basic/Targets/AMDGPU.cpp
===================================================================
--- clang/lib/Basic/Targets/AMDGPU.cpp
+++ clang/lib/Basic/Targets/AMDGPU.cpp
@@ -304,6 +304,15 @@
}
MaxAtomicPromoteWidth = MaxAtomicInlineWidth = 64;
+
+ SupportsOpenCLImages = true;
+
+ // CL2.0 features require flat instruction support. ROCm supports CL2.0, but
+ // Clover does not.
+ const bool SupportsCL2 = hasFlatAddressSpace() &&
+ (Triple.getOS() == llvm::Triple::AMDHSA ||
+ Triple.getOS() == llvm::Triple::AMDPAL);
+ MaxOpenCLSupportedVersion = SupportsCL2 ? 200 : 120;
}
void AMDGPUTargetInfo::adjust(LangOptions &Opts) {
Index: clang/include/clang/Basic/TargetInfo.h
===================================================================
--- clang/include/clang/Basic/TargetInfo.h
+++ clang/include/clang/Basic/TargetInfo.h
@@ -214,6 +214,12 @@
unsigned MaxOpenCLWorkGroupSize;
+ bool SupportsOpenCLImages = false;
+
+ /// Maximum supported device OpenCL version, corresponding to the values
+ /// expected for __OPENCL_VERSION__. If 0, the macro is not defined.
+ unsigned MaxOpenCLSupportedVersion = 0;
+
// TargetInfo Constructor. Default initializes all fields.
TargetInfo(const llvm::Triple &T);
@@ -667,6 +673,16 @@
unsigned getMaxOpenCLWorkGroupSize() const { return MaxOpenCLWorkGroupSize; }
+ /// Returns if if OpenCL images are supported.
+ bool hasOpenCLImageSupport() const {
+ return SupportsOpenCLImages;
+ }
+
+ /// Return value to report for __OPENCL_VERSION__
+ unsigned getMaxOpenCLSupportedVersion() const {
+ return MaxOpenCLSupportedVersion;
+ }
+
/// Return the alignment (in bits) of the thrown exception object. This is
/// only meaningful for targets that allocate C++ exceptions in a system
/// runtime, such as those using the Itanium C++ ABI.
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
