From: Craig Topper <craig.topper@gmail.com>
Date: Tue, 20 Aug 2013 07:09:39 +0000 (+0000)
Subject: Add AVX-512 feature flag and knl cpu to clang.
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=10c2c68e5610a58f3dbd2daed303b21e15a83720;p=clang

Add AVX-512 feature flag and knl cpu to clang.

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@188758 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Basic/Targets.cpp b/lib/Basic/Targets.cpp
index c4772723e9..02c5935f29 100644
--- a/lib/Basic/Targets.cpp
+++ b/lib/Basic/Targets.cpp
@@ -1570,7 +1570,7 @@ const TargetInfo::AddlRegName AddlRegNames[] = {
 // most of the implementation can be shared.
 class X86TargetInfo : public TargetInfo {
   enum X86SSEEnum {
-    NoSSE, SSE1, SSE2, SSE3, SSSE3, SSE41, SSE42, AVX, AVX2
+    NoSSE, SSE1, SSE2, SSE3, SSSE3, SSE41, SSE42, AVX, AVX2, AVX512
   } SSELevel;
   enum MMX3DNowEnum {
     NoMMX3DNow, MMX, AMD3DNow, AMD3DNowAthlon
@@ -1675,6 +1675,10 @@ class X86TargetInfo : public TargetInfo {
     CK_CoreAVX2,
     //@}
 
+    /// \name Knights Landing
+    /// Knights Landing processor.
+    CK_KNL,
+
     /// \name K6
     /// K6 architecture processors.
     //@{
@@ -1818,6 +1822,7 @@ public:
       .Case("corei7-avx", CK_Corei7AVX)
       .Case("core-avx-i", CK_CoreAVXi)
       .Case("core-avx2", CK_CoreAVX2)
+      .Case("knl", CK_KNL)
       .Case("k6", CK_K6)
       .Case("k6-2", CK_K6_2)
       .Case("k6-3", CK_K6_3)
@@ -1892,6 +1897,7 @@ public:
     case CK_Corei7AVX:
     case CK_CoreAVXi:
     case CK_CoreAVX2:
+    case CK_KNL:
     case CK_Athlon64:
     case CK_Athlon64SSE3:
     case CK_AthlonFX:
@@ -1941,6 +1947,7 @@ void X86TargetInfo::getDefaultFeatures(llvm::StringMap<bool> &Features) const {
   Features["pclmul"] = false;
   Features["avx"] = false;
   Features["avx2"] = false;
+  Features["avx512"] = false;
   Features["lzcnt"] = false;
   Features["rdrand"] = false;
   Features["bmi"] = false;
@@ -2024,6 +2031,18 @@ void X86TargetInfo::getDefaultFeatures(llvm::StringMap<bool> &Features) const {
     setFeatureEnabled(Features, "rtm", true);
     setFeatureEnabled(Features, "fma", true);
     break;
+  case CK_KNL:
+    setFeatureEnabled(Features, "avx512", true);
+    setFeatureEnabled(Features, "aes", true);
+    setFeatureEnabled(Features, "pclmul", true);
+    setFeatureEnabled(Features, "lzcnt", true);
+    setFeatureEnabled(Features, "rdrnd", true);
+    setFeatureEnabled(Features, "f16c", true);
+    setFeatureEnabled(Features, "bmi", true);
+    setFeatureEnabled(Features, "bmi2", true);
+    setFeatureEnabled(Features, "rtm", true);
+    setFeatureEnabled(Features, "fma", true);
+    break;
   case CK_K6:
   case CK_WinChipC6:
     setFeatureEnabled(Features, "mmx", true);
@@ -2149,6 +2168,11 @@ bool X86TargetInfo::setFeatureEnabled(llvm::StringMap<bool> &Features,
       Features["mmx"] = Features["sse"] = Features["sse2"] = Features["sse3"] =
         Features["ssse3"] = Features["sse41"] = Features["sse42"] =
         Features["popcnt"] = Features["avx"] = Features["avx2"] = true;
+    else if (Name == "avx512")
+      Features["mmx"] = Features["sse"] = Features["sse2"] = Features["sse3"] =
+        Features["ssse3"] = Features["sse41"] = Features["sse42"] =
+        Features["popcnt"] = Features["avx"] = Features["avx2"] =
+        Features["avx512"] = true;
     else if (Name == "fma")
       Features["mmx"] = Features["sse"] = Features["sse2"] = Features["sse3"] =
         Features["ssse3"] = Features["sse41"] = Features["sse42"] =
@@ -2191,28 +2215,29 @@ bool X86TargetInfo::setFeatureEnabled(llvm::StringMap<bool> &Features,
       Features["sse"] = Features["sse2"] = Features["sse3"] =
         Features["ssse3"] = Features["sse41"] = Features["sse42"] =
         Features["sse4a"] = Features["avx"] = Features["avx2"] =
-        Features["fma"] = Features["fma4"] = Features["aes"] =
-        Features["pclmul"] = Features["xop"] = false;
+        Features["avx512"] = Features["fma"] = Features["fma4"] =
+        Features["aes"] = Features["pclmul"] = Features["xop"] = false;
     else if (Name == "sse2")
       Features["sse2"] = Features["sse3"] = Features["ssse3"] =
         Features["sse41"] = Features["sse42"] = Features["sse4a"] =
-        Features["avx"] = Features["avx2"] = Features["fma"] =
-        Features["fma4"] = Features["aes"] = Features["pclmul"] =
-        Features["xop"] = false;
+        Features["avx"] = Features["avx2"] = Features["avx512"] =
+        Features["fma"] = Features["fma4"] = Features["aes"] =
+        Features["pclmul"] = Features["xop"] = false;
     else if (Name == "sse3")
       Features["sse3"] = Features["ssse3"] = Features["sse41"] =
         Features["sse42"] = Features["sse4a"] = Features["avx"] =
-        Features["avx2"] = Features["fma"] = Features["fma4"] =
-        Features["xop"] = false;
+        Features["avx2"] = Features["avx512"] = Features["fma"] =
+        Features["fma4"] = Features["xop"] = false;
     else if (Name == "ssse3")
       Features["ssse3"] = Features["sse41"] = Features["sse42"] =
-        Features["avx"] = Features["avx2"] = Features["fma"] = false;
+        Features["avx"] = Features["avx2"] = Features["avx512"] =
+        Features["fma"] = false;
     else if (Name == "sse4" || Name == "sse4.1")
       Features["sse41"] = Features["sse42"] = Features["avx"] =
-        Features["avx2"] = Features["fma"] = false;
+        Features["avx2"] = Features["avx512"] = Features["fma"] = false;
     else if (Name == "sse4.2")
       Features["sse42"] = Features["avx"] = Features["avx2"] =
-        Features["fma"] = false;
+        Features["avx512"] = Features["fma"] = false;
     else if (Name == "3dnow")
       Features["3dnow"] = Features["3dnowa"] = false;
     else if (Name == "3dnowa")
@@ -2222,10 +2247,12 @@ bool X86TargetInfo::setFeatureEnabled(llvm::StringMap<bool> &Features,
     else if (Name == "pclmul")
       Features["pclmul"] = false;
     else if (Name == "avx")
-      Features["avx"] = Features["avx2"] = Features["fma"] =
-        Features["fma4"] = Features["xop"] = false;
+      Features["avx"] = Features["avx2"] = Features["avx512"] =
+        Features["fma"] = Features["fma4"] = Features["xop"] = false;
     else if (Name == "avx2")
-      Features["avx2"] = false;
+      Features["avx2"] = Features["avx512"] = false;
+    else if (Name == "avx512")
+      Features["avx512"] = false;
     else if (Name == "fma")
       Features["fma"] = false;
     else if (Name == "sse4a")
@@ -2345,6 +2372,7 @@ void X86TargetInfo::HandleTargetFeatures(std::vector<std::string> &Features) {
 
     assert(Features[i][0] == '+' && "Invalid target feature!");
     X86SSEEnum Level = llvm::StringSwitch<X86SSEEnum>(Feature)
+      .Case("avx512", AVX512)
       .Case("avx2", AVX2)
       .Case("avx", AVX)
       .Case("sse42", SSE42)
@@ -2455,6 +2483,9 @@ void X86TargetInfo::getTargetDefines(const LangOptions &Opts,
   case CK_CoreAVX2:
     defineCPUMacros(Builder, "corei7");
     break;
+  case CK_KNL:
+    defineCPUMacros(Builder, "knl");
+    break;
   case CK_K6_2:
     Builder.defineMacro("__k6_2__");
     Builder.defineMacro("__tune_k6_2__");
@@ -2568,6 +2599,8 @@ void X86TargetInfo::getTargetDefines(const LangOptions &Opts,
 
   // Each case falls through to the previous one here.
   switch (SSELevel) {
+  case AVX512:
+    Builder.defineMacro("__AVX512__");
   case AVX2:
     Builder.defineMacro("__AVX2__");
   case AVX:
@@ -2592,6 +2625,7 @@ void X86TargetInfo::getTargetDefines(const LangOptions &Opts,
 
   if (Opts.MicrosoftExt && getTriple().getArch() == llvm::Triple::x86) {
     switch (SSELevel) {
+    case AVX512:
     case AVX2:
     case AVX:
     case SSE42:
@@ -2635,6 +2669,7 @@ bool X86TargetInfo::hasFeature(StringRef Feature) const {
       .Case("aes", HasAES)
       .Case("avx", SSELevel >= AVX)
       .Case("avx2", SSELevel >= AVX2)
+      .Case("avx512", SSELevel >= AVX512)
       .Case("bmi", HasBMI)
       .Case("bmi2", HasBMI2)
       .Case("fma", HasFMA)
diff --git a/test/Preprocessor/predefined-arch-macros.c b/test/Preprocessor/predefined-arch-macros.c
index 2944d0988e..803d752f2a 100644
--- a/test/Preprocessor/predefined-arch-macros.c
+++ b/test/Preprocessor/predefined-arch-macros.c
@@ -509,6 +509,7 @@
 // RUN:     -target i386-unknown-linux \
 // RUN:   | FileCheck %s -check-prefix=CHECK_CORE_AVX2_M32
 // CHECK_CORE_AVX2_M32: #define __AES__ 1
+// CHECK_CORE_AVX2_M32: #define __AVX2__ 1
 // CHECK_CORE_AVX2_M32: #define __AVX__ 1
 // CHECK_CORE_AVX2_M32: #define __BMI2__ 1
 // CHECK_CORE_AVX2_M32: #define __BMI__ 1
@@ -536,6 +537,7 @@
 // RUN:     -target i386-unknown-linux \
 // RUN:   | FileCheck %s -check-prefix=CHECK_CORE_AVX2_M64
 // CHECK_CORE_AVX2_M64: #define __AES__ 1
+// CHECK_CORE_AVX2_M64: #define __AVX2__ 1
 // CHECK_CORE_AVX2_M64: #define __AVX__ 1
 // CHECK_CORE_AVX2_M64: #define __BMI2__ 1
 // CHECK_CORE_AVX2_M64: #define __BMI__ 1
@@ -563,6 +565,68 @@
 // CHECK_CORE_AVX2_M64: #define __x86_64 1
 // CHECK_CORE_AVX2_M64: #define __x86_64__ 1
 //
+// RUN: %clang -march=knl -m32 -E -dM %s -o - 2>&1 \
+// RUN:     -target i386-unknown-linux \
+// RUN:   | FileCheck %s -check-prefix=CHECK_KNL_M32
+// CHECK_KNL_M32: #define __AES__ 1
+// CHECK_KNL_M32: #define __AVX2__ 1
+// CHECK_KNL_M32: #define __AVX512__ 1
+// CHECK_KNL_M32: #define __AVX__ 1
+// CHECK_KNL_M32: #define __BMI2__ 1
+// CHECK_KNL_M32: #define __BMI__ 1
+// CHECK_KNL_M32: #define __F16C__ 1
+// CHECK_KNL_M32: #define __FMA__ 1
+// CHECK_KNL_M32: #define __LZCNT__ 1
+// CHECK_KNL_M32: #define __MMX__ 1
+// CHECK_KNL_M32: #define __PCLMUL__ 1
+// CHECK_KNL_M32: #define __POPCNT__ 1
+// CHECK_KNL_M32: #define __RDRND__ 1
+// CHECK_KNL_M32: #define __RTM__ 1
+// CHECK_KNL_M32: #define __SSE2__ 1
+// CHECK_KNL_M32: #define __SSE3__ 1
+// CHECK_KNL_M32: #define __SSE4_1__ 1
+// CHECK_KNL_M32: #define __SSE4_2__ 1
+// CHECK_KNL_M32: #define __SSE__ 1
+// CHECK_KNL_M32: #define __SSSE3__ 1
+// CHECK_KNL_M32: #define __i386 1
+// CHECK_KNL_M32: #define __i386__ 1
+// CHECK_KNL_M32: #define __knl 1
+// CHECK_KNL_M32: #define __knl__ 1
+// CHECK_KNL_M32: #define __tune_knl__ 1
+// CHECK_KNL_M32: #define i386 1
+// RUN: %clang -march=knl -m64 -E -dM %s -o - 2>&1 \
+// RUN:     -target i386-unknown-linux \
+// RUN:   | FileCheck %s -check-prefix=CHECK_KNL_M64
+// CHECK_KNL_M64: #define __AES__ 1
+// CHECK_KNL_M64: #define __AVX2__ 1
+// CHECK_KNL_M64: #define __AVX512__ 1
+// CHECK_KNL_M64: #define __AVX__ 1
+// CHECK_KNL_M64: #define __BMI2__ 1
+// CHECK_KNL_M64: #define __BMI__ 1
+// CHECK_KNL_M64: #define __F16C__ 1
+// CHECK_KNL_M64: #define __FMA__ 1
+// CHECK_KNL_M64: #define __LZCNT__ 1
+// CHECK_KNL_M64: #define __MMX__ 1
+// CHECK_KNL_M64: #define __PCLMUL__ 1
+// CHECK_KNL_M64: #define __POPCNT__ 1
+// CHECK_KNL_M64: #define __RDRND__ 1
+// CHECK_KNL_M64: #define __RTM__ 1
+// CHECK_KNL_M64: #define __SSE2_MATH__ 1
+// CHECK_KNL_M64: #define __SSE2__ 1
+// CHECK_KNL_M64: #define __SSE3__ 1
+// CHECK_KNL_M64: #define __SSE4_1__ 1
+// CHECK_KNL_M64: #define __SSE4_2__ 1
+// CHECK_KNL_M64: #define __SSE_MATH__ 1
+// CHECK_KNL_M64: #define __SSE__ 1
+// CHECK_KNL_M64: #define __SSSE3__ 1
+// CHECK_KNL_M64: #define __amd64 1
+// CHECK_KNL_M64: #define __amd64__ 1
+// CHECK_KNL_M64: #define __knl 1
+// CHECK_KNL_M64: #define __knl__ 1
+// CHECK_KNL_M64: #define __tune_knl__ 1
+// CHECK_KNL_M64: #define __x86_64 1
+// CHECK_KNL_M64: #define __x86_64__ 1
+//
 // RUN: %clang -march=atom -m32 -E -dM %s -o - 2>&1 \
 // RUN:     -target i386-unknown-linux \
 // RUN:   | FileCheck %s -check-prefix=CHECK_ATOM_M32