tra created this revision.
tra added reviewers: yaxunl, jdoerfert.
Herald added subscribers: dexonsmith, bixia, hiraditya, jholewinski.
tra requested review of this revision.
Herald added projects: clang, LLVM.
The patch only plumbs through the option necessary for targeting sm_86 GPUs w/o
adding any new functionality.
Repository:
rG LLVM Github Monorepo
https://reviews.llvm.org/D95974
Files:
clang/include/clang/Basic/BuiltinsNVPTX.def
clang/include/clang/Basic/Cuda.h
clang/lib/Basic/Cuda.cpp
clang/lib/Basic/Targets/NVPTX.cpp
clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp
clang/lib/Driver/ToolChains/Cuda.cpp
llvm/lib/Target/NVPTX/NVPTX.td
Index: llvm/lib/Target/NVPTX/NVPTX.td
===================================================================
--- llvm/lib/Target/NVPTX/NVPTX.td
+++ llvm/lib/Target/NVPTX/NVPTX.td
@@ -57,6 +57,8 @@
"Target SM 7.5">;
def SM80 : SubtargetFeature<"sm_80", "SmVersion", "80",
"Target SM 8.0">;
+def SM86 : SubtargetFeature<"sm_86", "SmVersion", "86",
+ "Target SM 8.6">;
// PTX Versions
def PTX32 : SubtargetFeature<"ptx32", "PTXVersion", "32",
@@ -83,6 +85,10 @@
"Use PTX version 6.5">;
def PTX70 : SubtargetFeature<"ptx70", "PTXVersion", "70",
"Use PTX version 7.0">;
+def PTX71 : SubtargetFeature<"ptx71", "PTXVersion", "71",
+ "Use PTX version 7.1">;
+def PTX72 : SubtargetFeature<"ptx72", "PTXVersion", "72",
+ "Use PTX version 7.2">;
//===----------------------------------------------------------------------===//
// NVPTX supported processors.
@@ -107,6 +113,7 @@
def : Proc<"sm_72", [SM72, PTX61]>;
def : Proc<"sm_75", [SM75, PTX63]>;
def : Proc<"sm_80", [SM80, PTX70]>;
+def : Proc<"sm_86", [SM86, PTX71]>;
def NVPTXInstrInfo : InstrInfo {
}
Index: clang/lib/Driver/ToolChains/Cuda.cpp
===================================================================
--- clang/lib/Driver/ToolChains/Cuda.cpp
+++ clang/lib/Driver/ToolChains/Cuda.cpp
@@ -75,6 +75,8 @@
return CudaVersion::CUDA_102;
if (raw_version < 11010)
return CudaVersion::CUDA_110;
+ if (raw_version < 11020)
+ return CudaVersion::CUDA_111;
return CudaVersion::LATEST;
}
@@ -98,6 +100,7 @@
StartsWithWords(Input.ltrim(), {"#", "define", "CUDA_VERSION"})) {
uint32_t RawVersion;
Line->consumeInteger(10, RawVersion);
+ llvm::errs() << "raw = " << RawVersion << "\n";
return {"cuda.h: CUDA_VERSION=" + Twine(RawVersion).str() + ".",
getCudaVersion(RawVersion)};
}
@@ -724,6 +727,8 @@
CudaVersionStr = #CUDA_VER; \
PtxFeature = "+ptx" #PTX_VER; \
break;
+ CASE_CUDA_VERSION(112, 72);
+ CASE_CUDA_VERSION(111, 71);
CASE_CUDA_VERSION(110, 70);
CASE_CUDA_VERSION(102, 65);
CASE_CUDA_VERSION(101, 64);
Index: clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp
===================================================================
--- clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp
+++ clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp
@@ -4613,6 +4613,7 @@
case CudaArch::SM_72:
case CudaArch::SM_75:
case CudaArch::SM_80:
+ case CudaArch::SM_86:
case CudaArch::GFX600:
case CudaArch::GFX601:
case CudaArch::GFX602:
@@ -4680,6 +4681,7 @@
case CudaArch::SM_72:
case CudaArch::SM_75:
case CudaArch::SM_80:
+ case CudaArch::SM_86:
return {84, 32};
case CudaArch::GFX600:
case CudaArch::GFX601:
Index: clang/lib/Basic/Targets/NVPTX.cpp
===================================================================
--- clang/lib/Basic/Targets/NVPTX.cpp
+++ clang/lib/Basic/Targets/NVPTX.cpp
@@ -45,6 +45,8 @@
if (!Feature.startswith("+ptx"))
continue;
PTXVersion = llvm::StringSwitch<unsigned>(Feature)
+ .Case("+ptx72", 72)
+ .Case("+ptx71", 71)
.Case("+ptx70", 70)
.Case("+ptx65", 65)
.Case("+ptx64", 64)
@@ -246,6 +248,8 @@
return "750";
case CudaArch::SM_80:
return "800";
+ case CudaArch::SM_86:
+ return "860";
}
llvm_unreachable("unhandled CudaArch");
}();
Index: clang/lib/Basic/Cuda.cpp
===================================================================
--- clang/lib/Basic/Cuda.cpp
+++ clang/lib/Basic/Cuda.cpp
@@ -32,6 +32,10 @@
return "10.2";
case CudaVersion::CUDA_110:
return "11.0";
+ case CudaVersion::CUDA_111:
+ return "11.1";
+ case CudaVersion::CUDA_112:
+ return "11.2";
}
llvm_unreachable("invalid enum");
}
@@ -48,6 +52,8 @@
.Case("10.1", CudaVersion::CUDA_101)
.Case("10.2", CudaVersion::CUDA_102)
.Case("11.0", CudaVersion::CUDA_110)
+ .Case("11.1", CudaVersion::CUDA_111)
+ .Case("11.2", CudaVersion::CUDA_112)
.Default(CudaVersion::UNKNOWN);
}
@@ -71,7 +77,7 @@
SM(60), SM(61), SM(62), // Pascal
SM(70), SM(72), // Volta
SM(75), // Turing
- SM(80), // Ampere
+ SM(80), SM(86), // Ampere
GFX(600), // gfx600
GFX(601), // gfx601
GFX(602), // gfx602
@@ -164,6 +170,8 @@
return CudaVersion::CUDA_100;
case CudaArch::SM_80:
return CudaVersion::CUDA_110;
+ case CudaArch::SM_86:
+ return CudaVersion::CUDA_111;
default:
llvm_unreachable("invalid enum");
}
@@ -209,6 +217,10 @@
return CudaVersion::CUDA_102;
case 110:
return CudaVersion::CUDA_110;
+ case 111:
+ return CudaVersion::CUDA_111;
+ case 112:
+ return CudaVersion::CUDA_112;
default:
return CudaVersion::UNKNOWN;
}
Index: clang/include/clang/Basic/Cuda.h
===================================================================
--- clang/include/clang/Basic/Cuda.h
+++ clang/include/clang/Basic/Cuda.h
@@ -29,7 +29,9 @@
CUDA_101,
CUDA_102,
CUDA_110,
- LATEST = CUDA_110,
+ CUDA_111,
+ CUDA_112,
+ LATEST = CUDA_112,
LATEST_SUPPORTED = CUDA_101,
};
const char *CudaVersionToString(CudaVersion V);
@@ -55,6 +57,7 @@
SM_72,
SM_75,
SM_80,
+ SM_86,
GFX600,
GFX601,
GFX602,
Index: clang/include/clang/Basic/BuiltinsNVPTX.def
===================================================================
--- clang/include/clang/Basic/BuiltinsNVPTX.def
+++ clang/include/clang/Basic/BuiltinsNVPTX.def
@@ -21,7 +21,9 @@
#pragma push_macro("SM_72")
#pragma push_macro("SM_75")
#pragma push_macro("SM_80")
-#define SM_80 "sm_80"
+#pragma push_macro("SM_86")
+#define SM_86 "sm_86"
+#define SM_80 "sm_80|" SM_86
#define SM_75 "sm_75|" SM_80
#define SM_72 "sm_72|" SM_75
#define SM_70 "sm_70|" SM_72
@@ -35,7 +37,9 @@
#pragma push_macro("PTX64")
#pragma push_macro("PTX65")
#pragma push_macro("PTX70")
-#define PTX70 "ptx70"
+#pragma push_macro("PTX71")
+#define PTX71 "ptx71"
+#define PTX70 "ptx70|" PTX71
#define PTX65 "ptx65|" PTX70
#define PTX64 "ptx64|" PTX65
#define PTX63 "ptx63|" PTX64
@@ -728,9 +732,11 @@
#pragma pop_macro("SM_72")
#pragma pop_macro("SM_75")
#pragma pop_macro("SM_80")
+#pragma pop_macro("SM_86")
#pragma pop_macro("PTX60")
#pragma pop_macro("PTX61")
#pragma pop_macro("PTX63")
#pragma pop_macro("PTX64")
#pragma pop_macro("PTX65")
#pragma pop_macro("PTX70")
+#pragma pop_macro("PTX71")
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
