diff options
| author | Simon Pilgrim <llvm-dev@redking.me.uk> | 2019-02-01 15:31:01 +0000 |
|---|---|---|
| committer | Simon Pilgrim <llvm-dev@redking.me.uk> | 2019-02-01 15:31:01 +0000 |
| commit | 1a529f58f959255d954b48e9ee65db2dd3225a10 (patch) | |
| tree | 2a6e1e9ac5d3af38abd8c01a7cf01ffa0c91fd69 /llvm/test/CodeGen/X86 | |
| parent | 05a3f9978ea904860c4072d893dd5bcc5920eab0 (diff) | |
| download | bcm5719-llvm-1a529f58f959255d954b48e9ee65db2dd3225a10.tar.gz bcm5719-llvm-1a529f58f959255d954b48e9ee65db2dd3225a10.zip | |
[X86][AVX] Combine INSERT_SUBVECTOR(SRC0, BITCAST(SHUFFLE(EXTRACT_SUBVECTOR(SRC1)))
Enable peeking through one use bitcasts to the subvector shuffle.
This still depends on the subvector being the same scalar-size but D57514 has already helped with the more tricky patterns
llvm-svn: 352879
Diffstat (limited to 'llvm/test/CodeGen/X86')
| -rw-r--r-- | llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll | 25 |
1 files changed, 16 insertions, 9 deletions
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll b/llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll index 63c65d7abda..2eb33ff0af8 100644 --- a/llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll +++ b/llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll @@ -1553,12 +1553,19 @@ define <8 x i32> @shuffle_v8i32_08991abb(<8 x i32> %a, <8 x i32> %b) { ; AVX2-NEXT: vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7] ; AVX2-NEXT: retq ; -; AVX512VL-LABEL: shuffle_v8i32_08991abb: -; AVX512VL: # %bb.0: -; AVX512VL-NEXT: vpmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero -; AVX512VL-NEXT: vmovdqa {{.*#+}} ymm0 = [8,0,1,1,10,2,3,3] -; AVX512VL-NEXT: vpermi2d %ymm2, %ymm1, %ymm0 -; AVX512VL-NEXT: retq +; AVX512VL-SLOW-LABEL: shuffle_v8i32_08991abb: +; AVX512VL-SLOW: # %bb.0: +; AVX512VL-SLOW-NEXT: vpmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero +; AVX512VL-SLOW-NEXT: vmovdqa {{.*#+}} ymm0 = [8,0,1,1,10,2,3,3] +; AVX512VL-SLOW-NEXT: vpermi2d %ymm2, %ymm1, %ymm0 +; AVX512VL-SLOW-NEXT: retq +; +; AVX512VL-FAST-LABEL: shuffle_v8i32_08991abb: +; AVX512VL-FAST: # %bb.0: +; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} ymm2 = [8,0,1,1,9,2,3,3] +; AVX512VL-FAST-NEXT: vpermi2d %ymm0, %ymm1, %ymm2 +; AVX512VL-FAST-NEXT: vmovdqa %ymm2, %ymm0 +; AVX512VL-FAST-NEXT: retq %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 8, i32 9, i32 9, i32 1, i32 10, i32 11, i32 11> ret <8 x i32> %shuffle } @@ -1605,9 +1612,9 @@ define <8 x i32> @shuffle_v8i32_09ab1def(<8 x i32> %a, <8 x i32> %b) { ; ; AVX512VL-FAST-LABEL: shuffle_v8i32_09ab1def: ; AVX512VL-FAST: # %bb.0: -; AVX512VL-FAST-NEXT: vpmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero -; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} ymm0 = [8,1,2,3,10,5,6,7] -; AVX512VL-FAST-NEXT: vpermi2d %ymm2, %ymm1, %ymm0 +; AVX512VL-FAST-NEXT: vmovdqa {{.*#+}} ymm2 = [8,1,2,3,9,5,6,7] +; AVX512VL-FAST-NEXT: vpermi2d %ymm0, %ymm1, %ymm2 +; AVX512VL-FAST-NEXT: vmovdqa %ymm2, %ymm0 ; AVX512VL-FAST-NEXT: retq %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 1, i32 13, i32 14, i32 15> ret <8 x i32> %shuffle |
