[PATCH] [2/2] Add PRFCHW feature to a few AMD processors

Yunzhong Gao Yunzhong_Gao at playstation.sony.com
Tue Oct 15 15:41:29 PDT 2013 

  > Probably also need to check for hasPRFTCHW here
  > if (Subtarget->hasSSE1())
  >  setOperationAction(ISD::PREFETCH      , MVT::Other, Legal);

  ISD::PREFETCH can expand to other prefetch instructions such as prefetcht1 or prefetcht2, which are not covered by the PRFTCHW feature.

  I like your suggestion about splitting 3DNow! and PRFCHW features, and I updated my patch.

http://llvm-reviews.chandlerc.com/D1934

CHANGE SINCE LAST DIFF
  http://llvm-reviews.chandlerc.com/D1934?vs=4905&id=4937#toc

Files:
  lib/Basic/Targets.cpp
  test/Preprocessor/predefined-arch-macros.c
  test/Preprocessor/x86_target_features.c

Index: lib/Basic/Targets.cpp
===================================================================
--- lib/Basic/Targets.cpp
+++ lib/Basic/Targets.cpp
@@ -2120,6 +2120,7 @@
     setFeatureEnabledImpl(Features, "cx16", true);
     setFeatureEnabledImpl(Features, "lzcnt", true);
     setFeatureEnabledImpl(Features, "popcnt", true);
+    setFeatureEnabledImpl(Features, "prfchw", true);
     break;
   case CK_BTVER2:
     setFeatureEnabledImpl(Features, "avx", true);
@@ -2127,6 +2128,7 @@
     setFeatureEnabledImpl(Features, "lzcnt", true);
     setFeatureEnabledImpl(Features, "aes", true);
     setFeatureEnabledImpl(Features, "pclmul", true);
+    setFeatureEnabledImpl(Features, "prfchw", true);
     setFeatureEnabledImpl(Features, "bmi", true);
     setFeatureEnabledImpl(Features, "f16c", true);
     setFeatureEnabledImpl(Features, "cx16", true);
@@ -2136,6 +2138,7 @@
     setFeatureEnabledImpl(Features, "lzcnt", true);
     setFeatureEnabledImpl(Features, "aes", true);
     setFeatureEnabledImpl(Features, "pclmul", true);
+    setFeatureEnabledImpl(Features, "prfchw", true);
     setFeatureEnabledImpl(Features, "cx16", true);
     break;
   case CK_BDVER2:
@@ -2143,6 +2146,7 @@
     setFeatureEnabledImpl(Features, "lzcnt", true);
     setFeatureEnabledImpl(Features, "aes", true);
     setFeatureEnabledImpl(Features, "pclmul", true);
+    setFeatureEnabledImpl(Features, "prfchw", true);
     setFeatureEnabledImpl(Features, "bmi", true);
     setFeatureEnabledImpl(Features, "fma", true);
     setFeatureEnabledImpl(Features, "f16c", true);
@@ -2467,6 +2471,13 @@
     Features.push_back("+popcnt");
   }
 
+  // Enable prfchw if 3DNow! is enabled and prfchw is not explicitly disabled.
+  if (!HasPRFCHW && MMX3DNowLevel >= AMD3DNow &&
+      std::find(Features.begin(), Features.end(), "-prfchw") == Features.end()){
+    HasPRFCHW = true;
+    Features.push_back("+prfchw");
+  }
+
   // LLVM doesn't have a separate switch for fpmath, so only accept it if it
   // matches the selected sse level.
   if (FPMath == FP_SSE && SSELevel < SSE1) {
Index: test/Preprocessor/predefined-arch-macros.c
===================================================================
--- test/Preprocessor/predefined-arch-macros.c
+++ test/Preprocessor/predefined-arch-macros.c
@@ -1166,6 +1166,7 @@
 // CHECK_BTVER1_M32: #define __LZCNT__ 1
 // CHECK_BTVER1_M32: #define __MMX__ 1
 // CHECK_BTVER1_M32: #define __POPCNT__ 1
+// CHECK_BTVER1_M32: #define __PRFCHW__ 1
 // CHECK_BTVER1_M32: #define __SSE2_MATH__ 1
 // CHECK_BTVER1_M32: #define __SSE2__ 1
 // CHECK_BTVER1_M32: #define __SSE3__ 1
@@ -1186,6 +1187,7 @@
 // CHECK_BTVER1_M64: #define __LZCNT__ 1
 // CHECK_BTVER1_M64: #define __MMX__ 1
 // CHECK_BTVER1_M64: #define __POPCNT__ 1
+// CHECK_BTVER1_M64: #define __PRFCHW__ 1
 // CHECK_BTVER1_M64: #define __SSE2_MATH__ 1
 // CHECK_BTVER1_M64: #define __SSE2__ 1
 // CHECK_BTVER1_M64: #define __SSE3__ 1
@@ -1210,6 +1212,7 @@
 // CHECK_BTVER2_M32: #define __LZCNT__ 1
 // CHECK_BTVER2_M32: #define __MMX__ 1
 // CHECK_BTVER2_M32: #define __POPCNT__ 1
+// CHECK_BTVER2_M32: #define __PRFCHW__ 1
 // CHECK_BTVER2_M32: #define __SSE2_MATH__ 1
 // CHECK_BTVER2_M32: #define __SSE2__ 1
 // CHECK_BTVER2_M32: #define __SSE3__ 1
@@ -1232,6 +1235,7 @@
 // CHECK_BTVER2_M64: #define __LZCNT__ 1
 // CHECK_BTVER2_M64: #define __MMX__ 1
 // CHECK_BTVER2_M64: #define __POPCNT__ 1
+// CHECK_BTVER2_M64: #define __PRFCHW__ 1
 // CHECK_BTVER2_M64: #define __SSE2_MATH__ 1
 // CHECK_BTVER2_M64: #define __SSE2__ 1
 // CHECK_BTVER2_M64: #define __SSE3__ 1
@@ -1258,6 +1262,7 @@
 // CHECK_BDVER1_M32: #define __MMX__ 1
 // CHECK_BDVER1_M32: #define __PCLMUL__ 1
 // CHECK_BDVER1_M32: #define __POPCNT__ 1
+// CHECK_BDVER1_M32: #define __PRFCHW__ 1
 // CHECK_BDVER1_M32: #define __SSE2_MATH__ 1
 // CHECK_BDVER1_M32: #define __SSE2__ 1
 // CHECK_BDVER1_M32: #define __SSE3__ 1
@@ -1285,6 +1290,7 @@
 // CHECK_BDVER1_M64: #define __MMX__ 1
 // CHECK_BDVER1_M64: #define __PCLMUL__ 1
 // CHECK_BDVER1_M64: #define __POPCNT__ 1
+// CHECK_BDVER1_M64: #define __PRFCHW__ 1
 // CHECK_BDVER1_M64: #define __SSE2_MATH__ 1
 // CHECK_BDVER1_M64: #define __SSE2__ 1
 // CHECK_BDVER1_M64: #define __SSE3__ 1
@@ -1317,6 +1323,7 @@
 // CHECK_BDVER2_M32: #define __MMX__ 1
 // CHECK_BDVER2_M32: #define __PCLMUL__ 1
 // CHECK_BDVER2_M32: #define __POPCNT__ 1
+// CHECK_BDVER2_M32: #define __PRFCHW__ 1
 // CHECK_BDVER2_M32: #define __SSE2_MATH__ 1
 // CHECK_BDVER2_M32: #define __SSE2__ 1
 // CHECK_BDVER2_M32: #define __SSE3__ 1
@@ -1348,6 +1355,7 @@
 // CHECK_BDVER2_M64: #define __MMX__ 1
 // CHECK_BDVER2_M64: #define __PCLMUL__ 1
 // CHECK_BDVER2_M64: #define __POPCNT__ 1
+// CHECK_BDVER2_M64: #define __PRFCHW__ 1
 // CHECK_BDVER2_M64: #define __SSE2_MATH__ 1
 // CHECK_BDVER2_M64: #define __SSE2__ 1
 // CHECK_BDVER2_M64: #define __SSE3__ 1
Index: test/Preprocessor/x86_target_features.c
===================================================================
--- test/Preprocessor/x86_target_features.c
+++ test/Preprocessor/x86_target_features.c
@@ -215,3 +215,24 @@
 // RUN: %clang -target i386-unknown-unknown -march=pentiumpro -mcx16 -x c -E -dM -o - %s | FileCheck --check-prefix=MCX16 %s
 
 // MCX16: #define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16 1
+
+// RUN: %clang -target i386-unknown-unknown -march=atom -mprfchw -x c -E -dM -o - %s | FileCheck --check-prefix=PRFCHW %s
+
+// PRFCHW: #define __PRFCHW__ 1
+
+// RUN: %clang -target i386-unknown-unknown -march=btver2 -mno-prfchw -x c -E -dM -o - %s | FileCheck --check-prefix=NOPRFCHW %s
+
+// NOPRFCHW-NOT: #define __PRFCHW__ 1
+
+// RUN: %clang -target i386-unknown-unknown -march=atom -m3dnow -x c -E -dM -o - %s | FileCheck --check-prefix=3DNOWPRFCHW %s
+
+// 3DNOWPRFCHW: #define __PRFCHW__ 1
+
+// RUN: %clang -target i386-unknown-unknown -march=atom -mno-prfchw -m3dnow -x c -E -dM -o - %s | FileCheck --check-prefix=3DNOWNOPRFCHW %s
+
+// 3DNOWNOPRFCHW-NOT: #define __PRFCHW__ 1
+
+// RUN: %clang -target i386-unknown-unknown -march=atom -mprfchw -mno-3dnow -x c -E -dM -o - %s | FileCheck --check-prefix=NO3DNOWPRFCHW %s
+
+// NO3DNOWPRFCHW: #define __PRFCHW__ 1
+
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D1934.2.patch
Type: text/x-patch
Size: 6113 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20131015/1ea4573d/attachment.bin>

More information about the cfe-commits mailing list