[X86][SSE] Add support for combining target shuffles to PSLLDQ/PSRLDQ byte shifts

llvm-svn: 278502

8 files changed, 35 insertions, 43 deletions

diff --git a/llvm/test/CodeGen/X86/insertelement-zero.ll b/llvm/test/CodeGen/X86/insertelement-zero.ll
index d139b72f8ac..ac27bb7d8af 100644
--- a/llvm/test/CodeGen/X86/insertelement-zero.ll
+++ b/llvm/test/CodeGen/X86/insertelement-zero.ll
@@ -480,7 +480,7 @@ define <16 x i8> @insert_v16i8_z123456789ABZDEz(<16 x i8> %a) {
 ; SSSE3-NEXT:    pshufb {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
 ; SSSE3-NEXT:    por %xmm2, %xmm0
 ; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14],zero
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0]
+; SSSE3-NEXT:    pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0]
 ; SSSE3-NEXT:    por %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
@@ -560,7 +560,7 @@ define <32 x i8> @insert_v32i8_z123456789ABCDEzGHIJKLMNOPQRSTzz(<32 x i8> %a) {
 ; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,128]
 ; SSSE3-NEXT:    pshufb %xmm3, %xmm0
 ; SSSE3-NEXT:    movdqa %xmm2, %xmm4
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm4 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm4[0]
+; SSSE3-NEXT:    pslldq {{.*#+}} xmm4 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm4[0]
 ; SSSE3-NEXT:    por %xmm4, %xmm0
 ; SSSE3-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,6,7,8,9,10,11,12,13],zero,xmm1[15]
 ; SSSE3-NEXT:    pshufb {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm2[0],zero
diff --git a/llvm/test/CodeGen/X86/merge-consecutive-loads-256.ll b/llvm/test/CodeGen/X86/merge-consecutive-loads-256.ll
index dc268d9bdf8..254d712b760 100644
--- a/llvm/test/CodeGen/X86/merge-consecutive-loads-256.ll
+++ b/llvm/test/CodeGen/X86/merge-consecutive-loads-256.ll
@@ -253,26 +253,26 @@ define <4 x i64> @merge_4i64_i64_23zz(i64* %ptr) nounwind uwtable noinline ssp {
 define <8 x float> @merge_8f32_2f32_23z5(<2 x float>* %ptr) nounwind uwtable noinline ssp {
 ; AVX1-LABEL: merge_8f32_2f32_23z5:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vmovupd 16(%rdi), %xmm0
-; AVX1-NEXT:    vxorpd %xmm1, %xmm1, %xmm1
-; AVX1-NEXT:    vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX1-NEXT:    vmovups 16(%rdi), %xmm1
+; AVX1-NEXT:    vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: merge_8f32_2f32_23z5:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovupd 16(%rdi), %xmm0
-; AVX2-NEXT:    vxorpd %xmm1, %xmm1, %xmm1
-; AVX2-NEXT:    vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
-; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX2-NEXT:    vmovdqu 16(%rdi), %xmm1
+; AVX2-NEXT:    vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7]
+; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: merge_8f32_2f32_23z5:
 ; AVX512F:       # BB#0:
-; AVX512F-NEXT:    vmovups 16(%rdi), %xmm0
-; AVX512F-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
-; AVX512F-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX512F-NEXT:    vmovdqu 16(%rdi), %xmm1
+; AVX512F-NEXT:    vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7]
+; AVX512F-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX512F-NEXT:    retq
 ;
 ; X32-AVX-LABEL: merge_8f32_2f32_23z5:
diff --git a/llvm/test/CodeGen/X86/vector-half-conversions.ll b/llvm/test/CodeGen/X86/vector-half-conversions.ll
index 26db8fc5418..6fb698be968 100644
--- a/llvm/test/CodeGen/X86/vector-half-conversions.ll
+++ b/llvm/test/CodeGen/X86/vector-half-conversions.ll
@@ -3241,8 +3241,7 @@ define <8 x i16> @cvt_4f32_to_8i16_zero(<4 x float> %a0) nounwind {
 ; AVX512VL-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
 ; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
-; AVX512VL-NEXT:    vpxord %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
 ; AVX512VL-NEXT:    addq $24, %rsp
 ; AVX512VL-NEXT:    popq %rbx
 ; AVX512VL-NEXT:    popq %r14
@@ -4149,8 +4148,7 @@ define void @store_cvt_4f32_to_8i16_zero(<4 x float> %a0, <8 x i16>* %a1) nounwi
 ; AVX512VL-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
 ; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
-; AVX512VL-NEXT:    vpxord %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
 ; AVX512VL-NEXT:    vmovdqa32 %xmm0, (%r14)
 ; AVX512VL-NEXT:    addq $16, %rsp
 ; AVX512VL-NEXT:    popq %rbx
@@ -5138,8 +5136,7 @@ define <8 x i16> @cvt_4f64_to_8i16_zero(<4 x double> %a0) nounwind {
 ; AVX512VL-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
 ; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
-; AVX512VL-NEXT:    vpxord %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
 ; AVX512VL-NEXT:    addq $40, %rsp
 ; AVX512VL-NEXT:    popq %rbx
 ; AVX512VL-NEXT:    popq %r14
@@ -5942,8 +5939,7 @@ define void @store_cvt_4f64_to_8i16_zero(<4 x double> %a0, <8 x i16>* %a1) nounw
 ; AVX512VL-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
 ; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
-; AVX512VL-NEXT:    vpxord %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
 ; AVX512VL-NEXT:    vmovdqa32 %xmm0, (%r14)
 ; AVX512VL-NEXT:    addq $32, %rsp
 ; AVX512VL-NEXT:    popq %rbx
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-256-v32.ll b/llvm/test/CodeGen/X86/vector-shuffle-256-v32.ll
index 788bea1e921..539113813f7 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-256-v32.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-256-v32.ll
@@ -622,7 +622,7 @@ define <32 x i8> @shuffle_v32i8_31_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_
 ; AVX1-NEXT:    vmovd %eax, %xmm2
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm2
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
 ; AVX1-NEXT:    vpor %xmm0, %xmm2, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll b/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll
index a10ba6ccc41..f2d74dcf2e9 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll
@@ -280,7 +280,7 @@ define <8 x float> @combine_permps_as_permpd(<8 x float> %a) {
 define <32 x i8> @combine_pshufb_as_pslldq(<32 x i8> %a0) {
 ; CHECK-LABEL: combine_pshufb_as_pslldq:
 ; CHECK:       # BB#0:
-; CHECK-NEXT:    vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[0,1,2,3,4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,18,19,20,21]
+; CHECK-NEXT:    vpslldq {{.*#+}} ymm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[0,1,2,3,4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,18,19,20,21]
 ; CHECK-NEXT:    retq
   %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5>)
   ret <32 x i8> %res0
@@ -289,7 +289,7 @@ define <32 x i8> @combine_pshufb_as_pslldq(<32 x i8> %a0) {
 define <32 x i8> @combine_pshufb_as_psrldq(<32 x i8> %a0) {
 ; CHECK-LABEL: combine_pshufb_as_psrldq:
 ; CHECK:       # BB#0:
-; CHECK-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vpsrldq {{.*#+}} ymm0 = ymm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
 ; CHECK-NEXT:    retq
   %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 15, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 15, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128>)
   ret <32 x i8> %res0
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll b/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
index 2a27483c28b..d0ad07d6d07 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
@@ -187,12 +187,12 @@ define <16 x i8> @combine_pshufb_psrldq(<16 x i8> %a0) {
 define <16 x i8> @combine_pshufb_as_pslldq(<16 x i8> %a0) {
 ; SSE-LABEL: combine_pshufb_as_pslldq:
 ; SSE:       # BB#0:
-; SSE-NEXT:    pshufb {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
+; SSE-NEXT:    pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_pshufb_as_pslldq:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
+; AVX-NEXT:    vpslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5]
 ; AVX-NEXT:    retq
   %res0 = call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %a0, <16 x i8> <i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5>)
   ret <16 x i8> %res0
@@ -201,12 +201,12 @@ define <16 x i8> @combine_pshufb_as_pslldq(<16 x i8> %a0) {
 define <16 x i8> @combine_pshufb_as_psrldq(<16 x i8> %a0) {
 ; SSE-LABEL: combine_pshufb_as_psrldq:
 ; SSE:       # BB#0:
-; SSE-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_pshufb_as_psrldq:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; AVX-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
 ; AVX-NEXT:    retq
   %res0 = call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %a0, <16 x i8> <i8 15, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128>)
   ret <16 x i8> %res0
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-combining.ll b/llvm/test/CodeGen/X86/vector-shuffle-combining.ll
index 6170eda1971..128ead339f9 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-combining.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-combining.ll
@@ -607,16 +607,14 @@ define <4 x i32> @combine_bitwise_ops_test3c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ; SSE2:       # BB#0:
 ; SSE2-NEXT:    pxor %xmm1, %xmm0
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
-; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; SSE2-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test3c:
 ; SSSE3:       # BB#0:
 ; SSSE3-NEXT:    pxor %xmm1, %xmm0
 ; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,2]
-; SSSE3-NEXT:    pxor %xmm1, %xmm1
-; SSSE3-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; SSSE3-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test3c:
@@ -728,17 +726,15 @@ define <4 x i32> @combine_bitwise_ops_test6c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ; SSE2-LABEL: combine_bitwise_ops_test6c:
 ; SSE2:       # BB#0:
 ; SSE2-NEXT:    pxor %xmm1, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
-; SSE2-NEXT:    pxor %xmm0, %xmm0
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
+; SSE2-NEXT:    pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test6c:
 ; SSSE3:       # BB#0:
 ; SSSE3-NEXT:    pxor %xmm1, %xmm0
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,3,2,3]
-; SSSE3-NEXT:    pxor %xmm0, %xmm0
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,3,2,3]
+; SSSE3-NEXT:    pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test6c:
diff --git a/llvm/test/CodeGen/X86/vector-zext.ll b/llvm/test/CodeGen/X86/vector-zext.ll
index cb3583cdbd0..68cdc8498f9 100644
--- a/llvm/test/CodeGen/X86/vector-zext.ll
+++ b/llvm/test/CodeGen/X86/vector-zext.ll
@@ -2068,10 +2068,10 @@ define <4 x i64> @shuf_zext_4i32_to_4i64_offset1(<4 x i32> %A) nounwind uwtable
 ;
 ; AVX1-LABEL: shuf_zext_4i32_to_4i64_offset1:
 ; AVX1:       # BB#0: # %entry
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm1 = xmm0[3],zero,zero,zero
-; AVX1-NEXT:    vxorps %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vblendps {{.*#+}} xmm0 = xmm2[0,1],xmm0[2],xmm2[3]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0,1],xmm0[2],xmm1[3]
+; AVX1-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuf_zext_4i32_to_4i64_offset1: