diff options
Diffstat (limited to 'llvm/test/CodeGen/X86/vector-rotate-256.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/vector-rotate-256.ll | 33 |
1 files changed, 17 insertions, 16 deletions
diff --git a/llvm/test/CodeGen/X86/vector-rotate-256.ll b/llvm/test/CodeGen/X86/vector-rotate-256.ll index 5ae3e2f5d762..a183b0c94080 100644 --- a/llvm/test/CodeGen/X86/vector-rotate-256.ll +++ b/llvm/test/CodeGen/X86/vector-rotate-256.ll @@ -175,28 +175,29 @@ define <16 x i16> @var_rotate_v16i16(<16 x i16> %a, <16 x i16> %b) nounwind { ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 ; AVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15] ; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm2 -; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm4 = xmm2[4,4,5,5,6,6,7,7] -; AVX1-NEXT: vpslld $23, %xmm4, %xmm4 -; AVX1-NEXT: vbroadcastss {{.*#+}} xmm5 = [1065353216,1065353216,1065353216,1065353216] -; AVX1-NEXT: vpaddd %xmm5, %xmm4, %xmm4 -; AVX1-NEXT: vcvttps2dq %xmm4, %xmm4 +; AVX1-NEXT: vpxor %xmm4, %xmm4, %xmm4 +; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm5 = xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7] +; AVX1-NEXT: vpslld $23, %xmm5, %xmm5 +; AVX1-NEXT: vbroadcastss {{.*#+}} xmm6 = [1065353216,1065353216,1065353216,1065353216] +; AVX1-NEXT: vpaddd %xmm6, %xmm5, %xmm5 +; AVX1-NEXT: vcvttps2dq %xmm5, %xmm5 ; AVX1-NEXT: vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero ; AVX1-NEXT: vpslld $23, %xmm2, %xmm2 -; AVX1-NEXT: vpaddd %xmm5, %xmm2, %xmm2 +; AVX1-NEXT: vpaddd %xmm6, %xmm2, %xmm2 ; AVX1-NEXT: vcvttps2dq %xmm2, %xmm2 -; AVX1-NEXT: vpackusdw %xmm4, %xmm2, %xmm2 -; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm4 -; AVX1-NEXT: vpmulhuw %xmm2, %xmm4, %xmm6 -; AVX1-NEXT: vpmullw %xmm2, %xmm4, %xmm2 -; AVX1-NEXT: vpor %xmm6, %xmm2, %xmm2 +; AVX1-NEXT: vpackusdw %xmm5, %xmm2, %xmm2 +; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm5 +; AVX1-NEXT: vpmulhuw %xmm2, %xmm5, %xmm7 +; AVX1-NEXT: vpmullw %xmm2, %xmm5, %xmm2 +; AVX1-NEXT: vpor %xmm7, %xmm2, %xmm2 ; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm1 -; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm3 = xmm1[4,4,5,5,6,6,7,7] +; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm3 = xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7] ; AVX1-NEXT: vpslld $23, %xmm3, %xmm3 -; AVX1-NEXT: vpaddd %xmm5, %xmm3, %xmm3 +; AVX1-NEXT: vpaddd %xmm6, %xmm3, %xmm3 ; AVX1-NEXT: vcvttps2dq %xmm3, %xmm3 ; AVX1-NEXT: vpmovzxwd {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero ; AVX1-NEXT: vpslld $23, %xmm1, %xmm1 -; AVX1-NEXT: vpaddd %xmm5, %xmm1, %xmm1 +; AVX1-NEXT: vpaddd %xmm6, %xmm1, %xmm1 ; AVX1-NEXT: vcvttps2dq %xmm1, %xmm1 ; AVX1-NEXT: vpackusdw %xmm3, %xmm1, %xmm1 ; AVX1-NEXT: vpmulhuw %xmm1, %xmm0, %xmm3 @@ -895,10 +896,10 @@ define <8 x i32> @constant_rotate_v8i32(<8 x i32> %a) nounwind { ; AVX1-LABEL: constant_rotate_v8i32: ; AVX1: # %bb.0: ; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[1,1,3,3] -; AVX1-NEXT: vpmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 # [32,u,128,u] +; AVX1-NEXT: vpmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 # [32,32,128,128] ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2 ; AVX1-NEXT: vpshufd {{.*#+}} xmm3 = xmm2[1,1,3,3] -; AVX1-NEXT: vpmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3, %xmm3 # [512,u,2048,u] +; AVX1-NEXT: vpmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3, %xmm3 # [512,512,2048,2048] ; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm1, %ymm1 ; AVX1-NEXT: vpmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 # [16,32,64,128] ; AVX1-NEXT: vpmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2 # [256,512,1024,2048] |