[cfe-commits] r154906 - in /cfe/trunk: include/clang/Basic/BuiltinsX86.def lib/Headers/avx2intrin.h lib/Headers/avxintrin.h test/CodeGen/avx-shuffle-builtins.c test/CodeGen/avx2-builtins.c test/CodeGen/builtins-x86.c
Craig Topper
craig.topper at gmail.com
Mon Apr 16 22:16:57 PDT 2012
Author: ctopper
Date: Tue Apr 17 00:16:56 2012
New Revision: 154906
URL: http://llvm.org/viewvc/llvm-project?rev=154906&view=rev
Log:
Convert vperm2f128 and vperm2i128 intrinsics back to using llvm intrinsics. Unfortunately, these instructions have behavior that can't be modeled with shuffle vector.
Modified:
cfe/trunk/include/clang/Basic/BuiltinsX86.def
cfe/trunk/lib/Headers/avx2intrin.h
cfe/trunk/lib/Headers/avxintrin.h
cfe/trunk/test/CodeGen/avx-shuffle-builtins.c
cfe/trunk/test/CodeGen/avx2-builtins.c
cfe/trunk/test/CodeGen/builtins-x86.c
Modified: cfe/trunk/include/clang/Basic/BuiltinsX86.def
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/BuiltinsX86.def?rev=154906&r1=154905&r2=154906&view=diff
==============================================================================
--- cfe/trunk/include/clang/Basic/BuiltinsX86.def (original)
+++ cfe/trunk/include/clang/Basic/BuiltinsX86.def Tue Apr 17 00:16:56 2012
@@ -412,6 +412,9 @@
BUILTIN(__builtin_ia32_cvttpd2dq256, "V4iV4d", "")
BUILTIN(__builtin_ia32_cvtpd2dq256, "V4iV4d", "")
BUILTIN(__builtin_ia32_cvttps2dq256, "V8iV8f", "")
+BUILTIN(__builtin_ia32_vperm2f128_pd256, "V4dV4dV4dIc", "")
+BUILTIN(__builtin_ia32_vperm2f128_ps256, "V8fV8fV8fIc", "")
+BUILTIN(__builtin_ia32_vperm2f128_si256, "V8iV8iV8iIc", "")
BUILTIN(__builtin_ia32_vinsertf128_pd256, "V4dV4dV2dIc", "")
BUILTIN(__builtin_ia32_vinsertf128_ps256, "V8fV8fV4fIc", "")
BUILTIN(__builtin_ia32_vinsertf128_si256, "V8iV8iV4iIc", "")
@@ -561,6 +564,7 @@
BUILTIN(__builtin_ia32_pbroadcastq128, "V2LLiV2LLi", "")
BUILTIN(__builtin_ia32_permvarsi256, "V8iV8iV8i", "")
BUILTIN(__builtin_ia32_permvarsf256, "V8fV8fV8f", "")
+BUILTIN(__builtin_ia32_permti256, "V4LLiV4LLiV4LLiIc", "")
BUILTIN(__builtin_ia32_extract128i256, "V2LLiV4LLiIc", "")
BUILTIN(__builtin_ia32_insert128i256, "V4LLiV4LLiV2LLiIc", "")
BUILTIN(__builtin_ia32_maskloadd256, "V8iV8iC*V8i", "")
Modified: cfe/trunk/lib/Headers/avx2intrin.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avx2intrin.h?rev=154906&r1=154905&r2=154906&view=diff
==============================================================================
--- cfe/trunk/lib/Headers/avx2intrin.h (original)
+++ cfe/trunk/lib/Headers/avx2intrin.h Tue Apr 17 00:16:56 2012
@@ -841,11 +841,7 @@
#define _mm256_permute2x128_si256(V1, V2, M) __extension__ ({ \
__m256i __V1 = (V1); \
__m256i __V2 = (V2); \
- __builtin_shufflevector(__V1, __V2, \
- ((M) & 0x3) * 2, \
- ((M) & 0x3) * 2 + 1, \
- (((M) & 0x30) >> 4) * 2, \
- (((M) & 0x30) >> 4) * 2 + 1); })
+ (__m256i)__builtin_ia32_permti256(__V1, __V2, (M)); })
#define _mm256_extracti128_si256(A, O) __extension__ ({ \
__m256i __A = (A); \
Modified: cfe/trunk/lib/Headers/avxintrin.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avxintrin.h?rev=154906&r1=154905&r2=154906&view=diff
==============================================================================
--- cfe/trunk/lib/Headers/avxintrin.h (original)
+++ cfe/trunk/lib/Headers/avxintrin.h Tue Apr 17 00:16:56 2012
@@ -289,37 +289,17 @@
#define _mm256_permute2f128_pd(V1, V2, M) __extension__ ({ \
__m256d __V1 = (V1); \
__m256d __V2 = (V2); \
- (__m256d)__builtin_shufflevector((__v4df)__V1, (__v4df)__V2, \
- ((M) & 0x3) * 2, \
- ((M) & 0x3) * 2 + 1, \
- (((M) & 0x30) >> 4) * 2, \
- (((M) & 0x30) >> 4) * 2 + 1); })
+ (__m256d)__builtin_ia32_vperm2f128_pd256((__v4df)__V1, (__v4df)__V2, (M)); })
#define _mm256_permute2f128_ps(V1, V2, M) __extension__ ({ \
__m256 __V1 = (V1); \
__m256 __V2 = (V2); \
- (__m256)__builtin_shufflevector((__v8sf)__V1, (__v8sf)__V2, \
- ((M) & 0x3) * 4, \
- ((M) & 0x3) * 4 + 1, \
- ((M) & 0x3) * 4 + 2, \
- ((M) & 0x3) * 4 + 3, \
- (((M) & 0x30) >> 4) * 4, \
- (((M) & 0x30) >> 4) * 4 + 1, \
- (((M) & 0x30) >> 4) * 4 + 2, \
- (((M) & 0x30) >> 4) * 4 + 3); })
+ (__m256)__builtin_ia32_vperm2f128_ps256((__v8sf)__V1, (__v8sf)__V2, (M)); })
#define _mm256_permute2f128_si256(V1, V2, M) __extension__ ({ \
__m256i __V1 = (V1); \
__m256i __V2 = (V2); \
- (__m256i)__builtin_shufflevector((__v8si)__V1, (__v8si)__V2, \
- ((M) & 0x3) * 4, \
- ((M) & 0x3) * 4 + 1, \
- ((M) & 0x3) * 4 + 2, \
- ((M) & 0x3) * 4 + 3, \
- (((M) & 0x30) >> 4) * 4, \
- (((M) & 0x30) >> 4) * 4 + 1, \
- (((M) & 0x30) >> 4) * 4 + 2, \
- (((M) & 0x30) >> 4) * 4 + 3); })
+ (__m256i)__builtin_ia32_vperm2f128_si256((__v8si)__V1, (__v8si)__V2, (M)); })
/* Vector Blend */
#define _mm256_blend_pd(V1, V2, M) __extension__ ({ \
Modified: cfe/trunk/test/CodeGen/avx-shuffle-builtins.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/avx-shuffle-builtins.c?rev=154906&r1=154905&r2=154906&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/avx-shuffle-builtins.c (original)
+++ cfe/trunk/test/CodeGen/avx-shuffle-builtins.c Tue Apr 17 00:16:56 2012
@@ -48,18 +48,18 @@
__m256d test_mm256_permute2f128_pd(__m256d a, __m256d b) {
// Check if the mask is correct
- // CHECK: shufflevector{{.*}}<i32 2, i32 3, i32 6, i32 7>
+ // CHECK: @llvm.x86.avx.vperm2f128.pd.256
return _mm256_permute2f128_pd(a, b, 0x31);
}
__m256 test_mm256_permute2f128_ps(__m256 a, __m256 b) {
// Check if the mask is correct
- // CHECK: shufflevector{{.*}}<i32 12, i32 13, i32 14, i32 15, i32 4, i32 5, i32 6, i32 7>
+ // CHECK: @llvm.x86.avx.vperm2f128.ps.256
return _mm256_permute2f128_ps(a, b, 0x13);
}
__m256i test_mm256_permute2f128_si256(__m256i a, __m256i b) {
// Check if the mask is correct
- // CHECK: shufflevector{{.*}}<i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
+ // CHECK: @llvm.x86.avx.vperm2f128.si.256
return _mm256_permute2f128_si256(a, b, 0x20);
}
Modified: cfe/trunk/test/CodeGen/avx2-builtins.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/avx2-builtins.c?rev=154906&r1=154905&r2=154906&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/avx2-builtins.c (original)
+++ cfe/trunk/test/CodeGen/avx2-builtins.c Tue Apr 17 00:16:56 2012
@@ -677,7 +677,7 @@
}
__m256i test_mm256_permute2x128_si256(__m256i a, __m256i b) {
- // CHECK: shufflevector{{.*}}<i32 2, i32 3, i32 6, i32 7>
+ // CHECK: @llvm.x86.avx2.vperm2i128
return _mm256_permute2x128_si256(a, b, 0x31);
}
Modified: cfe/trunk/test/CodeGen/builtins-x86.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/builtins-x86.c?rev=154906&r1=154905&r2=154906&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/builtins-x86.c (original)
+++ cfe/trunk/test/CodeGen/builtins-x86.c Tue Apr 17 00:16:56 2012
@@ -414,6 +414,9 @@
tmp_V4i = __builtin_ia32_cvttpd2dq256(tmp_V4d);
tmp_V4i = __builtin_ia32_cvtpd2dq256(tmp_V4d);
tmp_V8i = __builtin_ia32_cvttps2dq256(tmp_V8f);
+ tmp_V4d = __builtin_ia32_vperm2f128_pd256(tmp_V4d, tmp_V4d, 0x7);
+ tmp_V8f = __builtin_ia32_vperm2f128_ps256(tmp_V8f, tmp_V8f, 0x7);
+ tmp_V8i = __builtin_ia32_vperm2f128_si256(tmp_V8i, tmp_V8i, 0x7);
tmp_V4d = __builtin_ia32_vinsertf128_pd256(tmp_V4d, tmp_V2d, 0x7);
tmp_V8f = __builtin_ia32_vinsertf128_ps256(tmp_V8f, tmp_V4f, 0x7);
tmp_V8i = __builtin_ia32_vinsertf128_si256(tmp_V8i, tmp_V4i, 0x7);
More information about the cfe-commits
mailing list