Emit predefined macro for wavefront size for amdgcn

[yaxunl]

Summary

Also fix the issue of multiple -m[no-]wavefrontsize64 options to make the last one wins.

[yaxunl]

fix typo

[arsenm]

clang/lib/Basic/Targets/AMDGPU.h

421

Why is this not redundant with the features check?

clang/test/Driver/amdgpu-macros.cl

351

Macros should be all caps

[yaxunl]

clang/lib/Basic/Targets/AMDGPU.h

421

You mean the above check for target ID features? It only checks features in target ID, i.e. xnack and sramecc.

clang/test/Driver/amdgpu-macros.cl

351

will change

[yaxunl]

capitalize macro

[yaxunl]

revised by Matt's comments.

[yaxunl]

ping

[arsenm]

clang/test/Driver/amdgpu-macros.cl

380–381

Can you also check with the flags reversed, -mno-wavefrontsize64 -mwavefrontsize64

[yaxunl]

Add test and fix multiple -m[no-]wavefrontsize64 issue.

[arsenm]

clang/lib/Driver/ToolChains/AMDGPU.cpp

396–397

Why isn't this using hasFlag?
e.g. like

DriverArgs.hasFlag(options::OPT_fcuda_flush_denormals_to_zero,
                           options::OPT_fno_cuda_flush_denormals_to_zero,
                           getDefaultDenormsAreZeroForTarget(Kind)))

[yaxunl]

clang/lib/Driver/ToolChains/AMDGPU.cpp

396–397

hasFlag always return true or false, but here we have 3 cases : no arg, last arg is wave64, last arg is no-wave64.

[arsenm]

clang/lib/Driver/ToolChains/AMDGPU.cpp

396–397

hasFlag considers all of these. The features only really need to add +wavefrontsze64 or not based on true or false, the rest is just noise

[yaxunl]

clang/lib/Driver/ToolChains/AMDGPU.cpp

396–397

I have a simpler solution. will update

[yaxunl]

simpler code for handling multiple wave64 options

[yaxunl]

simplifies wavefrontsize64 target feature

clang/lib/Basic/Targets/AMDGPU.h

enum AddrSpace {
     Constant = 4,
     Private = 5
   };
   static const LangASMap AMDGPUDefIsGenMap;
   static const LangASMap AMDGPUDefIsPrivMap;
 
   llvm::AMDGPU::GPUKind GPUKind;
   unsigned GPUFeatures;
+  unsigned WavefrontSize;
 
   /// Target ID is device name followed by optional feature name postfixed
   /// by plus or minus sign delimitted by colon, e.g. gfx908:xnack+:sram-ecc-.
   /// If the target ID contains feature+, map it to true.
   /// If the target ID contains feature-, map it to false.
   /// If the target ID does not contain a feature (default), do not map it.
   llvm::StringMap<bool> OffloadArchFeatures;
   std::string TargetID;
public:
   // Record offload arch features since they are needed for defining the
   // pre-defined macros.
   bool handleTargetFeatures(std::vector<std::string> &Features,
                             DiagnosticsEngine &Diags) override {
     auto TargetIDFeatures =
         getAllPossibleTargetIDFeatures(getTriple(), getArchNameAMDGCN(GPUKind));
     llvm::for_each(Features, [&](const auto &F) {
       assert(F.front() == '+' || F.front() == '-');
+      if (F == "+wavefrontsize64")
+        WavefrontSize = 64;
       bool IsOn = F.front() == '+';
       StringRef Name = StringRef(F).drop_front();
       if (llvm::find(TargetIDFeatures, Name) == TargetIDFeatures.end())
         return;
       assert(OffloadArchFeatures.find(Name) == OffloadArchFeatures.end());
       OffloadArchFeatures[Name] = IsOn;
     });
     return true;
   }
 
   Optional<std::string> getTargetID() const override {
     if (!isAMDGCN(getTriple()))
       return llvm::None;
     // When -target-cpu is not set, we assume generic code that it is valid
     // for all GPU and use an empty string as target ID to represent that.
     if (GPUKind == llvm::AMDGPU::GK_NONE)
       return std::string("");

clang/lib/Basic/Targets/AMDGPU.cpp

AMDGPUTargetInfo::AMDGPUTargetInfo(const llvm::Triple &Triple,
   GridValues = llvm::omp::AMDGPUGpuGridValues;
 
   setAddressSpaceMap(Triple.getOS() == llvm::Triple::Mesa3D ||
                      !isAMDGCN(Triple));
   UseAddrSpaceMapMangling = true;
 
   HasLegalHalfType = true;
   HasFloat16 = true;
+  WavefrontSize = GPUFeatures & llvm::AMDGPU::FEATURE_WAVE32 ? 32 : 64;
 
   // Set pointer width and alignment for target address space 0.
   PointerWidth = PointerAlign = DataLayout->getPointerSizeInBits();
   if (getMaxPointerWidth() == 64) {
     LongWidth = LongAlign = 64;
     SizeType = UnsignedLong;
     PtrDiffType = SignedLong;
     IntPtrType = SignedLong;
void AMDGPUTargetInfo::getTargetDefines(const LangOptions &Opts,
   if (hasFastFMAF())
     Builder.defineMacro("FP_FAST_FMAF");
   if (hasLDEXPF())
     Builder.defineMacro("__HAS_LDEXPF__");
   if (hasFP64())
     Builder.defineMacro("__HAS_FP64__");
   if (hasFastFMA())
     Builder.defineMacro("FP_FAST_FMA");
+
+  Builder.defineMacro("__AMDGCN_WAVEFRONT_SIZE", Twine(WavefrontSize));
 }
 
 void AMDGPUTargetInfo::setAuxTarget(const TargetInfo *Aux) {
   assert(HalfFormat == Aux->HalfFormat);
   assert(FloatFormat == Aux->FloatFormat);
   assert(DoubleFormat == Aux->DoubleFormat);
 
   // On x86_64 long double is 80-bit extended precision format, which is

clang/lib/Driver/ToolChains/AMDGPU.cpp

if (OptionalGpuArch) {
         if (Pos == FeatureMap.end())
           continue;
         Features.push_back(Args.MakeArgStringRef(
             (Twine(Pos->second ? "+" : "-") + Feature).str()));
       }
     }
   }
 
-  if (Args.getLastArg(options::OPT_mwavefrontsize64)) {
-    Features.push_back("-wavefrontsize16");
-    Features.push_back("-wavefrontsize32");
+  if (Args.hasFlag(options::OPT_mwavefrontsize64,
+                   options::OPT_mno_wavefrontsize64, false))
     Features.push_back("+wavefrontsize64");
-  }
-  if (Args.getLastArg(options::OPT_mno_wavefrontsize64)) {
-    Features.push_back("-wavefrontsize16");
-    Features.push_back("+wavefrontsize32");
-    Features.push_back("-wavefrontsize64");
-  }
 
   handleTargetFeaturesGroup(
     Args, Features, options::OPT_m_amdgpu_Features_Group);
 }
 
 /// AMDGPU Toolchain
 AMDGPUToolChain::AMDGPUToolChain(const Driver &D, const llvm::Triple &Triple,
                                  const ArgList &Args)
     : Generic_ELF(D, Triple, Args),
       OptionsDefault({{options::OPT_O, "3"},

clang/test/Driver/amdgpu-features.c

 
 // RUN: %clang -### -target amdgcn-mesa3d -mcpu=gfx908:sram-ecc+ %s 2>&1 | FileCheck --check-prefix=SRAM-ECC %s
 // SRAM-ECC: "-target-feature" "+sram-ecc"
 
 // RUN: %clang -### -target amdgcn-amdhsa -mcpu=gfx908:sram-ecc- %s 2>&1 | FileCheck --check-prefix=NO-SRAM-ECC %s
 // NO-SRAM-ECC: "-target-feature" "-sram-ecc"
 
 // RUN: %clang -### -target amdgcn-amdpal -mcpu=gfx1010 -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefix=WAVE64 %s
-// WAVE64: "-target-feature" "-wavefrontsize16" "-target-feature" "-wavefrontsize32" "-target-feature" "+wavefrontsize64"
+// RUN: %clang -### -target amdgcn-amdpal -mcpu=gfx1010 -mno-wavefrontsize64 -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefix=WAVE64 %s
+// WAVE64: "-target-feature" "+wavefrontsize64"
+// WAVE64-NOT: {{".*wavefrontsize16"}}
+// WAVE64-NOT: {{".*wavefrontsize32"}}
 
 // RUN: %clang -### -target amdgcn -mcpu=gfx1010 -mno-wavefrontsize64 %s 2>&1 | FileCheck --check-prefix=NO-WAVE64 %s
-// NO-WAVE64: "-target-feature" "-wavefrontsize16" "-target-feature" "+wavefrontsize32" "-target-feature" "-wavefrontsize64"
+// RUN: %clang -### -target amdgcn -mcpu=gfx1010 -mwavefrontsize64 -mno-wavefrontsize64 %s 2>&1 | FileCheck --check-prefix=NO-WAVE64 %s
+// NO-WAVE64-NOT: {{".*wavefrontsize16"}}
+// NO-WAVE64-NOT: {{".*wavefrontsize32"}}
+// NO-WAVE64-NOT: {{".*wavefrontsize64"}}
 
 // RUN: %clang -### -target amdgcn -mcpu=gfx1010 -mcumode %s 2>&1 | FileCheck --check-prefix=CUMODE %s
 // CUMODE: "-target-feature" "+cumode"
 
 // RUN: %clang -### -target amdgcn -mcpu=gfx1010 -mno-cumode %s 2>&1 | FileCheck --check-prefix=NO-CUMODE %s
 // NO-CUMODE: "-target-feature" "-cumode"

clang/test/Driver/amdgpu-macros.cl

 // GFX904-DAG: #define __amdgcn_processor__ "gfx904"
 // GFX906-DAG: #define __amdgcn_processor__ "gfx906"
 // GFX908-DAG: #define __amdgcn_processor__ "gfx908"
 // GFX909-DAG: #define __amdgcn_processor__ "gfx909"
 // GFX1010-DAG: #define __amdgcn_processor__ "gfx1010"
 // GFX1011-DAG: #define __amdgcn_processor__ "gfx1011"
 // GFX1012-DAG: #define __amdgcn_processor__ "gfx1012"
 // GFX1030-DAG: #define __amdgcn_processor__ "gfx1030"
-// GFX1031-DAG: #define __amdgcn_processor__ "gfx1031"
+// GFX1031-DAG: #define __amdgcn_processor__ "gfx1031"
+
+// GFX600-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX601-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX700-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX701-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX702-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX703-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX704-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX801-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX802-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX803-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX810-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX900-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX902-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX904-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX906-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX908-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX909-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// GFX1010-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32
+// GFX1011-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32
+// GFX1012-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32
+// GFX1030-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32
+// GFX1031-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32
+
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx906 -mwavefrontsize64 \
+// RUN:   %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx1010 -mwavefrontsize64 \
+// RUN:   %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx906 -mwavefrontsize64 \
+// RUN:   -mno-wavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx1010 -mwavefrontsize64 \
+// RUN:   -mno-wavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE32 %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx906 -mno-wavefrontsize64 \
+// RUN:   -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx1010 -mno-wavefrontsize64 \
+// RUN:   -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// WAVE64-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// WAVE32-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32

clang/test/Driver/hip-macros.hip

+// RUN: %clang -E -dM --offload-arch=gfx906 -mwavefrontsize64 \
+// RUN:   --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// RUN: %clang -E -dM --offload-arch=gfx1010 -mwavefrontsize64 \
+// RUN:   --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// RUN: %clang -E -dM --offload-arch=gfx906 -mwavefrontsize64 \
+// RUN:   --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   -mno-wavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// RUN: %clang -E -dM --offload-arch=gfx1010 -mwavefrontsize64 \
+// RUN:   --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   -mno-wavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE32 %s
+// RUN: %clang -E -dM --offload-arch=gfx906 -mno-wavefrontsize64 \
+// RUN:   --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// RUN: %clang -E -dM --offload-arch=gfx1010 -mno-wavefrontsize64 \
+// RUN:   --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
+// WAVE64-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
+// WAVE32-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32

clang/test/Driver/hip-toolchain-features.hip

 // ALL3: {{.*}}clang{{.*}}"-target-feature" "+sram-ecc" "-target-feature" "+xnack"
 // NOALL3: {{.*}}clang{{.*}}"-target-feature" "-sram-ecc" "-target-feature" "-xnack"
 
 // RUN: %clang -### -target x86_64-linux-gnu -fgpu-rdc -nogpulib \
 // RUN:   --cuda-gpu-arch=gfx1010 %s \
 // RUN:   -mcumode -mcumode -mno-cumode -mwavefrontsize64 -mcumode \
 // RUN:   -mwavefrontsize64 -mno-wavefrontsize64 2>&1 \
 // RUN:   | FileCheck %s -check-prefix=DUP
-// DUP: {{.*}}clang{{.*}} "-target-feature" "-wavefrontsize16"
-// DUP-SAME: "-target-feature" "+wavefrontsize32"
-// DUP-SAME: "-target-feature" "-wavefrontsize64"
-// DUP-SAME: "-target-feature" "+cumode"
-// DUP: {{.*}}lld{{.*}} "-plugin-opt=-mattr=-wavefrontsize16,+wavefrontsize32,-wavefrontsize64,+cumode"
+// DUP: {{.*}}clang{{.*}} "-target-feature" "+cumode"
+// DUP-NOT: "-target-feature" "{{.*}}wavefrontsize16"
+// DUP-NOT: "-target-feature" "{{.*}}wavefrontsize32"
+// DUP-NOT: "-target-feature" "{{.*}}wavefrontsize64"
+// DUP: {{.*}}lld{{.*}} "-plugin-opt=-mattr=+cumode"
+
+// RUN: %clang -### -target x86_64-linux-gnu -fgpu-rdc -nogpulib \
+// RUN:   --cuda-gpu-arch=gfx1010 %s \
+// RUN:   -mno-wavefrontsize64 -mwavefrontsize64 2>&1 \
+// RUN:   | FileCheck %s -check-prefix=WAVE64
+// WAVE64: {{.*}}clang{{.*}} "-target-feature" "+wavefrontsize64"
+// WAVE64-NOT: "-target-feature" "{{.*}}wavefrontsize16"
+// WAVE64-NOT: "-target-feature" "{{.*}}wavefrontsize32"
+// WAVE64: {{.*}}lld{{.*}} "-plugin-opt=-mattr=+wavefrontsize64"