1 files changed, 29 insertions, 15 deletions
diff --git a/llvm/test/CodeGen/X86/combine-sdiv.ll b/llvm/test/CodeGen/X86/combine-sdiv.ll
index 68b102eb0bc..3a3ee90d4dd 100644
--- a/llvm/test/CodeGen/X86/combine-sdiv.ll
+++ b/llvm/test/CodeGen/X86/combine-sdiv.ll
@@ -1580,7 +1580,7 @@ define <2 x i64> @combine_vec_sdiv_by_pow2b_v2i64(<2 x i64> %x) {
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [9223372036854775808,2305843009213693952]
 ; AVX2-NEXT:    vpxor %xmm2, %xmm1, %xmm1
 ; AVX2-NEXT:    vpsubq %xmm2, %xmm1, %xmm1
-; AVX2-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: combine_vec_sdiv_by_pow2b_v2i64:
@@ -1593,7 +1593,7 @@ define <2 x i64> @combine_vec_sdiv_by_pow2b_v2i64(<2 x i64> %x) {
 ; AVX512F-NEXT:    vpsrlvq {{.*}}(%rip), %xmm2, %xmm2
 ; AVX512F-NEXT:    vpaddq %xmm2, %xmm0, %xmm2
 ; AVX512F-NEXT:    vpsravq %zmm1, %zmm2, %zmm1
-; AVX512F-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX512F-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
@@ -1606,7 +1606,7 @@ define <2 x i64> @combine_vec_sdiv_by_pow2b_v2i64(<2 x i64> %x) {
 ; AVX512BW-NEXT:    vmovq %rax, %xmm2
 ; AVX512BW-NEXT:    vpslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,xmm2[0,1,2,3,4,5,6,7]
 ; AVX512BW-NEXT:    vpsravq %xmm2, %xmm1, %xmm1
-; AVX512BW-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX512BW-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
 ; AVX512BW-NEXT:    retq
 ;
 ; XOP-LABEL: combine_vec_sdiv_by_pow2b_v2i64:
@@ -2498,7 +2498,7 @@ define <8 x i16> @combine_vec_sdiv_nonuniform2(<8 x i16> %x) {
 ; AVX2-NEXT:    vpmulhw {{.*}}(%rip), %xmm0, %xmm0
 ; AVX2-NEXT:    vpsraw $1, %xmm0, %xmm1
 ; AVX2-NEXT:    vpsraw $2, %xmm0, %xmm2
-; AVX2-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3]
 ; AVX2-NEXT:    vpsrlw $15, %xmm0, %xmm0
 ; AVX2-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    retq
@@ -2508,7 +2508,7 @@ define <8 x i16> @combine_vec_sdiv_nonuniform2(<8 x i16> %x) {
 ; AVX512F-NEXT:    vpmulhw {{.*}}(%rip), %xmm0, %xmm0
 ; AVX512F-NEXT:    vpsraw $1, %xmm0, %xmm1
 ; AVX512F-NEXT:    vpsraw $2, %xmm0, %xmm2
-; AVX512F-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
+; AVX512F-NEXT:    vpblendd {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3]
 ; AVX512F-NEXT:    vpsrlw $15, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
 ; AVX512F-NEXT:    retq
@@ -2580,7 +2580,7 @@ define <8 x i16> @combine_vec_sdiv_nonuniform3(<8 x i16> %x) {
 ; AVX2-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vpsraw $8, %xmm0, %xmm1
 ; AVX2-NEXT:    vpsraw $4, %xmm0, %xmm2
-; AVX2-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3]
 ; AVX2-NEXT:    vpsrlw $15, %xmm0, %xmm0
 ; AVX2-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    retq
@@ -2591,7 +2591,7 @@ define <8 x i16> @combine_vec_sdiv_nonuniform3(<8 x i16> %x) {
 ; AVX512F-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
 ; AVX512F-NEXT:    vpsraw $8, %xmm0, %xmm1
 ; AVX512F-NEXT:    vpsraw $4, %xmm0, %xmm2
-; AVX512F-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
+; AVX512F-NEXT:    vpblendd {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3]
 ; AVX512F-NEXT:    vpsrlw $15, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
 ; AVX512F-NEXT:    retq
@@ -2665,7 +2665,7 @@ define <8 x i16> @combine_vec_sdiv_nonuniform4(<8 x i16> %x) {
 ; AVX2-NEXT:    vpsubw %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vpsraw $8, %xmm0, %xmm1
 ; AVX2-NEXT:    vpsraw $4, %xmm0, %xmm2
-; AVX2-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3]
 ; AVX2-NEXT:    vpsrlw $15, %xmm0, %xmm0
 ; AVX2-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    retq
@@ -2676,7 +2676,7 @@ define <8 x i16> @combine_vec_sdiv_nonuniform4(<8 x i16> %x) {
 ; AVX512F-NEXT:    vpsubw %xmm0, %xmm1, %xmm0
 ; AVX512F-NEXT:    vpsraw $8, %xmm0, %xmm1
 ; AVX512F-NEXT:    vpsraw $4, %xmm0, %xmm2
-; AVX512F-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
+; AVX512F-NEXT:    vpblendd {{.*#+}} xmm1 = xmm2[0,1],xmm1[2,3]
 ; AVX512F-NEXT:    vpsrlw $15, %xmm0, %xmm0
 ; AVX512F-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
 ; AVX512F-NEXT:    retq
@@ -2949,12 +2949,26 @@ define <8 x i16> @combine_vec_sdiv_nonuniform7(<8 x i16> %x) {
 ; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: combine_vec_sdiv_nonuniform7:
-; AVX:       # %bb.0:
-; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX-NEXT:    vpsubw %xmm0, %xmm1, %xmm1
-; AVX-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
-; AVX-NEXT:    retq
+; AVX1-LABEL: combine_vec_sdiv_nonuniform7:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsubw %xmm0, %xmm1, %xmm1
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2ORLATER-LABEL: combine_vec_sdiv_nonuniform7:
+; AVX2ORLATER:       # %bb.0:
+; AVX2ORLATER-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2ORLATER-NEXT:    vpsubw %xmm0, %xmm1, %xmm1
+; AVX2ORLATER-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3]
+; AVX2ORLATER-NEXT:    retq
+;
+; XOP-LABEL: combine_vec_sdiv_nonuniform7:
+; XOP:       # %bb.0:
+; XOP-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; XOP-NEXT:    vpsubw %xmm0, %xmm1, %xmm1
+; XOP-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
+; XOP-NEXT:    retq
   %1 = sdiv <8 x i16> %x, <i16 -1, i16 -1, i16 -1, i16 -1, i16 1, i16 1, i16 1, i16 1>
   ret <8 x i16> %1
 }