[cfe-commits] r147219 - in /cfe/trunk: include/clang/Basic/BuiltinsX86.def lib/Headers/avx2intrin.h test/CodeGen/avx2-builtins.c

[Craig Topper]

Author: ctopper
Date: Fri Dec 23 02:31:16 2011
New Revision: 147219

URL: http://llvm.org/viewvc/llvm-project?rev=147219&view=rev
Log:
Add AVX2 multiply intrinsics.

Modified:
    cfe/trunk/include/clang/Basic/BuiltinsX86.def
    cfe/trunk/lib/Headers/avx2intrin.h
    cfe/trunk/test/CodeGen/avx2-builtins.c

Modified: cfe/trunk/include/clang/Basic/BuiltinsX86.def
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/BuiltinsX86.def?rev=147219&r1=147218&r2=147219&view=diff
==============================================================================
--- cfe/trunk/include/clang/Basic/BuiltinsX86.def (original)
+++ cfe/trunk/include/clang/Basic/BuiltinsX86.def Fri Dec 23 02:31:16 2011
@@ -527,5 +527,10 @@
 BUILTIN(__builtin_ia32_pmovzxwd256, "V8iV8s", "")
 BUILTIN(__builtin_ia32_pmovzxwq256, "V4LLiV8s", "")
 BUILTIN(__builtin_ia32_pmovzxdq256, "V4LLiV4i", "")
+BUILTIN(__builtin_ia32_pmuldq256, "V4LLiV8iV8i", "")
+BUILTIN(__builtin_ia32_pmulhrsw256, "V16sV16sV16s", "")
+BUILTIN(__builtin_ia32_pmulhuw256, "V16sV16sV16s", "")
+BUILTIN(__builtin_ia32_pmulhw256, "V16sV16sV16s", "")
+BUILTIN(__builtin_ia32_pmuludq256, "V4LLiV8iV8i", "")
 
 #undef BUILTIN

Modified: cfe/trunk/lib/Headers/avx2intrin.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avx2intrin.h?rev=147219&r1=147218&r2=147219&view=diff
==============================================================================
--- cfe/trunk/lib/Headers/avx2intrin.h (original)
+++ cfe/trunk/lib/Headers/avx2intrin.h Fri Dec 23 02:31:16 2011
@@ -405,6 +405,48 @@
   return (__m256i)__builtin_ia32_pmovzxdq256((__v4si)__V);
 }
 
+static __inline__  __m256i __attribute__((__always_inline__, __nodebug__))
+_mm256_mul_epi32(__m256i a, __m256i b)
+{
+  return (__m256i)__builtin_ia32_pmuldq256((__v8si)a, (__v8si)b);
+}
+
+static __inline__ __m256i __attribute__((__always_inline__, __nodebug__))
+_mm256_mulhrs_epi16(__m256i a, __m256i b)
+{
+  return (__m256i)__builtin_ia32_pmulhrsw256((__v16hi)a, (__v16hi)b);
+}
+
+static __inline__ __m256i __attribute__((__always_inline__, __nodebug__))
+_mm256_mulhi_epu16(__m256i a, __m256i b)
+{
+  return (__m256i)__builtin_ia32_pmulhuw256((__v16hi)a, (__v16hi)b);
+}
+
+static __inline__ __m256i __attribute__((__always_inline__, __nodebug__))
+_mm256_mulhi_epi16(__m256i a, __m256i b)
+{
+  return (__m256i)__builtin_ia32_pmulhw256((__v16hi)a, (__v16hi)b);
+}
+
+static __inline__ __m256i __attribute__((__always_inline__, __nodebug__))
+_mm256_mullo_epi16(__m256i a, __m256i b)
+{
+  return (__m256i)((__v16hi)a * (__v16hi)b);
+}
+
+static __inline__  __m256i __attribute__((__always_inline__, __nodebug__))
+_mm256_mullo_epi32 (__m256i a, __m256i b)
+{
+  return (__m256i)((__v8si)a * (__v8si)b);
+}
+
+static __inline__ __m256i __attribute__((__always_inline__, __nodebug__))
+_mm256_mul_epu32(__m256i a, __m256i b)
+{
+  return __builtin_ia32_pmuludq256((__v8si)a, (__v8si)b);
+}
+
 static __inline__ __m256i __attribute__((__always_inline__, __nodebug__))
 _mm256_or_si256(__m256i a, __m256i b)
 {

Modified: cfe/trunk/test/CodeGen/avx2-builtins.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/avx2-builtins.c?rev=147219&r1=147218&r2=147219&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/avx2-builtins.c (original)
+++ cfe/trunk/test/CodeGen/avx2-builtins.c Fri Dec 23 02:31:16 2011
@@ -380,3 +380,38 @@
   // CHECK: @llvm.x86.avx2.pmovzxdq
   return _mm256_cvtepu32_epi64(a);
 }
+
+__m256 test_mm256_mul_epi32(__m256 a, __m256 b) {
+  // CHECK: @llvm.x86.avx2.pmul.dq
+  return _mm256_mul_epi32(a, b);
+}
+
+__m256 test_mm256_mulhrs_epi16(__m256 a, __m256 b) {
+  // CHECK: @llvm.x86.avx2.pmul.hr.sw
+  return _mm256_mulhrs_epi16(a, b);
+}
+
+__m256 test_mm256_mulhi_epu16(__m256 a, __m256 b) {
+  // CHECK: @llvm.x86.avx2.pmulhu.w
+  return _mm256_mulhi_epu16(a, b);
+}
+
+__m256 test_mm256_mulhi_epi16(__m256 a, __m256 b) {
+  // CHECK: @llvm.x86.avx2.pmulh.w
+  return _mm256_mulhi_epi16(a, b);
+}
+
+__m256 test_mm256_mullo_epi16(__m256 a, __m256 b) {
+  // CHECK: mul <16 x i16>
+  return _mm256_mullo_epi16(a, b);
+}
+
+__m256 test_mm256_mullo_epi32(__m256 a, __m256 b) {
+  // CHECK: mul <8 x i32>
+  return _mm256_mullo_epi32(a, b);
+}
+
+__m256 test_mm256_mul_epu32(__m256 a, __m256 b) {
+  // CHECK: @llvm.x86.avx2.pmulu.dq
+  return _mm256_mul_epu32(a, b);
+}