diff options
Diffstat (limited to 'llvm/test/CodeGen/X86/vector-sext.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/vector-sext.ll | 30 |
1 files changed, 12 insertions, 18 deletions
diff --git a/llvm/test/CodeGen/X86/vector-sext.ll b/llvm/test/CodeGen/X86/vector-sext.ll index d46514a6dc7..c458fb54629 100644 --- a/llvm/test/CodeGen/X86/vector-sext.ll +++ b/llvm/test/CodeGen/X86/vector-sext.ll @@ -449,7 +449,7 @@ define <4 x i64> @sext_16i8_to_4i64(<16 x i8> %A) nounwind uwtable readnone ssp ; SSSE3-NEXT: psrad $31, %xmm2 ; SSSE3-NEXT: psrad $24, %xmm0 ; SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1] -; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = xmm1[u,u,u,2,u,u,u,3,u,u,u],zero,xmm1[u,u,u],zero +; SSSE3-NEXT: pshufb {{.*#+}} xmm1 = xmm1[u,u,u,2,u,u,u,3,u,u,u,u,u,u,u,u] ; SSSE3-NEXT: movdqa %xmm1, %xmm2 ; SSSE3-NEXT: psrad $31, %xmm2 ; SSSE3-NEXT: psrad $24, %xmm1 @@ -530,7 +530,7 @@ define <8 x i64> @sext_16i8_to_8i64(<16 x i8> %A) nounwind uwtable readnone ssp ; ; SSSE3-LABEL: sext_16i8_to_8i64: ; SSSE3: # %bb.0: # %entry -; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = <u,u,u,2,u,u,u,3,u,u,u,255,u,u,u,255> +; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = <u,u,u,2,u,u,u,3,u,u,u,u,u,u,u,u> ; SSSE3-NEXT: punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3],xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7] ; SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm0[1,1,2,3] ; SSSE3-NEXT: movdqa %xmm0, %xmm1 @@ -836,8 +836,7 @@ define <4 x i64> @sext_8i16_to_4i64(<8 x i16> %A) nounwind uwtable readnone ssp ; SSE2-NEXT: psrad $31, %xmm1 ; SSE2-NEXT: psrad $16, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3] -; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] +; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm0[0,2,2,3,4,5,6,7] ; SSE2-NEXT: movdqa %xmm1, %xmm0 ; SSE2-NEXT: psrad $31, %xmm0 ; SSE2-NEXT: psrad $16, %xmm1 @@ -852,8 +851,7 @@ define <4 x i64> @sext_8i16_to_4i64(<8 x i16> %A) nounwind uwtable readnone ssp ; SSSE3-NEXT: psrad $31, %xmm1 ; SSSE3-NEXT: psrad $16, %xmm2 ; SSSE3-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3] -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] +; SSSE3-NEXT: pshuflw {{.*#+}} xmm1 = xmm0[0,2,2,3,4,5,6,7] ; SSSE3-NEXT: movdqa %xmm1, %xmm0 ; SSSE3-NEXT: psrad $31, %xmm0 ; SSSE3-NEXT: psrad $16, %xmm1 @@ -913,14 +911,13 @@ define <8 x i64> @sext_8i16_to_8i64(<8 x i16> %A) nounwind uwtable readnone ssp ; SSE2-NEXT: psrad $31, %xmm1 ; SSE2-NEXT: psrad $16, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3] -; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3] +; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm0[0,2,2,3,4,5,6,7] ; SSE2-NEXT: movdqa %xmm1, %xmm3 ; SSE2-NEXT: psrad $31, %xmm3 ; SSE2-NEXT: psrad $16, %xmm1 ; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3] -; SSE2-NEXT: punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1] +; SSE2-NEXT: pshuflw {{.*#+}} xmm3 = xmm0[0,2,2,3,4,5,6,7] ; SSE2-NEXT: movdqa %xmm3, %xmm0 ; SSE2-NEXT: psrad $31, %xmm0 ; SSE2-NEXT: psrad $16, %xmm3 @@ -940,14 +937,13 @@ define <8 x i64> @sext_8i16_to_8i64(<8 x i16> %A) nounwind uwtable readnone ssp ; SSSE3-NEXT: psrad $31, %xmm1 ; SSSE3-NEXT: psrad $16, %xmm2 ; SSSE3-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] -; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3] -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3] +; SSSE3-NEXT: pshuflw {{.*#+}} xmm1 = xmm0[0,2,2,3,4,5,6,7] ; SSSE3-NEXT: movdqa %xmm1, %xmm3 ; SSSE3-NEXT: psrad $31, %xmm3 ; SSSE3-NEXT: psrad $16, %xmm1 ; SSSE3-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1] -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3] -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3] +; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1] +; SSSE3-NEXT: pshuflw {{.*#+}} xmm3 = xmm0[0,2,2,3,4,5,6,7] ; SSSE3-NEXT: movdqa %xmm3, %xmm0 ; SSSE3-NEXT: psrad $31, %xmm0 ; SSSE3-NEXT: psrad $16, %xmm3 @@ -1295,8 +1291,7 @@ define <2 x i64> @load_sext_2i8_to_2i64(<2 x i8> *%ptr) { ; SSSE3: # %bb.0: # %entry ; SSSE3-NEXT: movzwl (%rdi), %eax ; SSSE3-NEXT: movd %eax, %xmm0 -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3] +; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[u,u,u,0,u,u,u,1,u,u,u,u,u,u,u,u] ; SSSE3-NEXT: movdqa %xmm0, %xmm1 ; SSSE3-NEXT: psrad $31, %xmm1 ; SSSE3-NEXT: psrad $24, %xmm0 @@ -5070,8 +5065,7 @@ define <2 x i32> @sext_2i8_to_2i32(<2 x i8>* %addr) { ; SSSE3: # %bb.0: ; SSSE3-NEXT: movzwl (%rdi), %eax ; SSSE3-NEXT: movd %eax, %xmm0 -; SSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7] -; SSSE3-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3] +; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[u,u,u,0,u,u,u,1,u,u,u,u,u,u,u,u] ; SSSE3-NEXT: psrad $24, %xmm0 ; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,1,3] ; SSSE3-NEXT: paddq %xmm0, %xmm0 |
