[X86][SSE] Fold extract_subvector(vselect(x,y,z),0) -> vselect(extract_subvector...

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@364136 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index b21f8fa25cdc84773f0f88024b481049017958fe..bb2545a24db5263a14ad9d9f91e5f9b721fa69a4 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -43623,6 +43623,16 @@ static SDValue combineExtractSubvector(SDNode *N, SelectionDAG &DAG,
       unsigned ExtOp = getOpcode_EXTEND_VECTOR_INREG(InOpcode);
       return DAG.getNode(ExtOp, SDLoc(N), VT, InVec.getOperand(0));
     }
+    if (InOpcode == ISD::VSELECT &&
+        InVec.getOperand(0).getValueType().is256BitVector() &&
+        InVec.getOperand(1).getValueType().is256BitVector() &&
+        InVec.getOperand(2).getValueType().is256BitVector()) {
+      SDLoc DL(N);
+      SDValue Ext0 = extractSubVector(InVec.getOperand(0), 0, DAG, DL, 128);
+      SDValue Ext1 = extractSubVector(InVec.getOperand(1), 0, DAG, DL, 128);
+      SDValue Ext2 = extractSubVector(InVec.getOperand(2), 0, DAG, DL, 128);
+      return DAG.getNode(InOpcode, DL, VT, Ext0, Ext1, Ext2);
+    }
   }
 
   return SDValue();

diff --git a/test/CodeGen/X86/horizontal-reduce-smax.ll b/test/CodeGen/X86/horizontal-reduce-smax.ll
index ce3210c477cdb1d200b132dfa26da213f627fe65..ec831ac9b9f6d60694def8a0c87ef7208594a8fa 100644 (file)
--- a/test/CodeGen/X86/horizontal-reduce-smax.ll
+++ b/test/CodeGen/X86/horizontal-reduce-smax.ll
@@ -545,7 +545,7 @@ define i64 @test_reduce_v4i64(<4 x i64> %a0) {
 ; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; X64-AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; X64-AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; X64-AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    vmovq %xmm0, %rax
 ; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq
@@ -1256,7 +1256,7 @@ define i64 @test_reduce_v8i64(<8 x i64> %a0) {
 ; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; X64-AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; X64-AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; X64-AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    vmovq %xmm0, %rax
 ; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/horizontal-reduce-smin.ll b/test/CodeGen/X86/horizontal-reduce-smin.ll
index eaae04e5c9489cd84839e36a84c02c36959cc055..1adb3368767d3dd5d6d8d467b5e4e15eac6a8442 100644 (file)
--- a/test/CodeGen/X86/horizontal-reduce-smin.ll
+++ b/test/CodeGen/X86/horizontal-reduce-smin.ll
@@ -549,7 +549,7 @@ define i64 @test_reduce_v4i64(<4 x i64> %a0) {
 ; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; X64-AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; X64-AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; X64-AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    vmovq %xmm0, %rax
 ; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq
@@ -1260,7 +1260,7 @@ define i64 @test_reduce_v8i64(<8 x i64> %a0) {
 ; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; X64-AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; X64-AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; X64-AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    vmovq %xmm0, %rax
 ; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/horizontal-reduce-umax.ll b/test/CodeGen/X86/horizontal-reduce-umax.ll
index 34df7ac33de464ddd74d041f47168c51bf1da2da..b676b2acb7752461e64cda0e8258f0dd4d0f8247 100644 (file)
--- a/test/CodeGen/X86/horizontal-reduce-umax.ll
+++ b/test/CodeGen/X86/horizontal-reduce-umax.ll
@@ -654,7 +654,7 @@ define i64 @test_reduce_v4i64(<4 x i64> %a0) {
 ; X64-AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; X64-AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; X64-AVX2-NEXT:    vpcmpgtq %ymm2, %ymm3, %ymm2
-; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; X64-AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    vmovq %xmm0, %rax
 ; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq
@@ -1434,7 +1434,7 @@ define i64 @test_reduce_v8i64(<8 x i64> %a0) {
 ; X64-AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; X64-AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; X64-AVX2-NEXT:    vpcmpgtq %ymm2, %ymm3, %ymm2
-; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; X64-AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    vmovq %xmm0, %rax
 ; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/horizontal-reduce-umin.ll b/test/CodeGen/X86/horizontal-reduce-umin.ll
index 2b21d27b048f05146367837d8cd1a7c2f9445777..8c524aa9269205db960db4247ccc45ce5bfdf104 100644 (file)
--- a/test/CodeGen/X86/horizontal-reduce-umin.ll
+++ b/test/CodeGen/X86/horizontal-reduce-umin.ll
@@ -594,7 +594,7 @@ define i64 @test_reduce_v4i64(<4 x i64> %a0) {
 ; X64-AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; X64-AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; X64-AVX2-NEXT:    vpcmpgtq %ymm3, %ymm2, %ymm2
-; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; X64-AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    vmovq %xmm0, %rax
 ; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq
@@ -1338,7 +1338,7 @@ define i64 @test_reduce_v8i64(<8 x i64> %a0) {
 ; X64-AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; X64-AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; X64-AVX2-NEXT:    vpcmpgtq %ymm3, %ymm2, %ymm2
-; X64-AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; X64-AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    vmovq %xmm0, %rax
 ; X64-AVX2-NEXT:    vzeroupper
 ; X64-AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/var-permute-128.ll b/test/CodeGen/X86/var-permute-128.ll
index 50cedfd2a292dc6a1fa169d4f1b607f41a29495f..49bbfa9d8cf1ab6fc858a2eb3492003176135fc6 100644 (file)
--- a/test/CodeGen/X86/var-permute-128.ll
+++ b/test/CodeGen/X86/var-permute-128.ll
@@ -1031,8 +1031,7 @@ define <16 x i8> @var_shuffle_v16i8_from_v32i8_v16i8(<32 x i8> %v, <16 x i8> %in
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
 ; AVX2-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpcmpgtb {{.*}}(%rip), %ymm1, %ymm1
-; AVX2-NEXT:    vpblendvb %ymm1, %ymm0, %ymm2, %ymm0
-; AVX2-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
+; AVX2-NEXT:    vpblendvb %xmm1, %xmm0, %xmm2, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
@@ -1043,8 +1042,7 @@ define <16 x i8> @var_shuffle_v16i8_from_v32i8_v16i8(<32 x i8> %v, <16 x i8> %in
 ; AVX512-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
 ; AVX512-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpcmpgtb {{.*}}(%rip), %ymm1, %ymm1
-; AVX512-NEXT:    vpblendvb %ymm1, %ymm0, %ymm2, %ymm0
-; AVX512-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
+; AVX512-NEXT:    vpblendvb %xmm1, %xmm0, %xmm2, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 ;

diff --git a/test/CodeGen/X86/var-permute-256.ll b/test/CodeGen/X86/var-permute-256.ll
index 5d5ed467e85dc53f6d688ecfea32e0710e395dc3..d1594eddf4444b4436f9ec478b5bcf6171f4fd20 100644 (file)
--- a/test/CodeGen/X86/var-permute-256.ll
+++ b/test/CodeGen/X86/var-permute-256.ll
@@ -1126,8 +1126,7 @@ define <4 x i32> @var_shuffle_v4i32_from_v8i32(<8 x i32> %v, <4 x i32> %indices)
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    vpermilps %ymm1, %ymm0, %ymm0
 ; AVX1-NEXT:    vpcmpgtd {{.*}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vblendvps %ymm1, %ymm2, %ymm0, %ymm0
-; AVX1-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
+; AVX1-NEXT:    vblendvps %xmm1, %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;

diff --git a/test/CodeGen/X86/vector-reduce-smax-widen.ll b/test/CodeGen/X86/vector-reduce-smax-widen.ll
index 447c17a366f504bebfd06ce15033fb889f7ac7c7..e1fe08ca92e8a734520cbfa8d8f14e2f6c788972 100644 (file)
--- a/test/CodeGen/X86/vector-reduce-smax-widen.ll
+++ b/test/CodeGen/X86/vector-reduce-smax-widen.ll
@@ -165,7 +165,7 @@ define i64 @test_v4i64(<4 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -342,7 +342,7 @@ define i64 @test_v8i64(<8 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -644,7 +644,7 @@ define i64 @test_v16i64(<16 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/vector-reduce-smax.ll b/test/CodeGen/X86/vector-reduce-smax.ll
index 1b15929bc19120297de6f7ce4b5e74f3874871f9..7cb0da94f95813ab4c02dba8b4a37a18ea6b6c47 100644 (file)
--- a/test/CodeGen/X86/vector-reduce-smax.ll
+++ b/test/CodeGen/X86/vector-reduce-smax.ll
@@ -165,7 +165,7 @@ define i64 @test_v4i64(<4 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -342,7 +342,7 @@ define i64 @test_v8i64(<8 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -644,7 +644,7 @@ define i64 @test_v16i64(<16 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/vector-reduce-smin-widen.ll b/test/CodeGen/X86/vector-reduce-smin-widen.ll
index 52e3b94036f0757a55de5a6824a08e06e3899fcd..25e4d0ddf8f79124519f8746e76ceb8ee40cb558 100644 (file)
--- a/test/CodeGen/X86/vector-reduce-smin-widen.ll
+++ b/test/CodeGen/X86/vector-reduce-smin-widen.ll
@@ -164,7 +164,7 @@ define i64 @test_v4i64(<4 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -341,7 +341,7 @@ define i64 @test_v8i64(<8 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -643,7 +643,7 @@ define i64 @test_v16i64(<16 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/vector-reduce-smin.ll b/test/CodeGen/X86/vector-reduce-smin.ll
index 9625f4c8579171fb0ae5b52f8121b632950145dd..b7c05a9eff95742def7d72bf9a12d51ce990cff9 100644 (file)
--- a/test/CodeGen/X86/vector-reduce-smin.ll
+++ b/test/CodeGen/X86/vector-reduce-smin.ll
@@ -164,7 +164,7 @@ define i64 @test_v4i64(<4 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -341,7 +341,7 @@ define i64 @test_v8i64(<8 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -643,7 +643,7 @@ define i64 @test_v16i64(<16 x i64> %a0) {
 ; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/vector-reduce-umax-widen.ll b/test/CodeGen/X86/vector-reduce-umax-widen.ll
index 7b3d4b4c1effbeb703b111075290fcf042bb4758..4e1cc5f23dfbcb2c84215bfe13efdb4c8fc157e1 100644 (file)
--- a/test/CodeGen/X86/vector-reduce-umax-widen.ll
+++ b/test/CodeGen/X86/vector-reduce-umax-widen.ll
@@ -178,7 +178,7 @@ define i64 @test_v4i64(<4 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpgtq %ymm2, %ymm3, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -371,7 +371,7 @@ define i64 @test_v8i64(<8 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpgtq %ymm2, %ymm3, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -701,7 +701,7 @@ define i64 @test_v16i64(<16 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm4, %ymm0, %ymm2
 ; AVX2-NEXT:    vxorpd %ymm4, %ymm1, %ymm3
 ; AVX2-NEXT:    vpcmpgtq %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/vector-reduce-umax.ll b/test/CodeGen/X86/vector-reduce-umax.ll
index f92153e3fa8f9e05af1dd7f798886c85e36eb0d0..e577232beebcd4e145667689ae666dcfe145b416 100644 (file)
--- a/test/CodeGen/X86/vector-reduce-umax.ll
+++ b/test/CodeGen/X86/vector-reduce-umax.ll
@@ -178,7 +178,7 @@ define i64 @test_v4i64(<4 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpgtq %ymm2, %ymm3, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -371,7 +371,7 @@ define i64 @test_v8i64(<8 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpgtq %ymm2, %ymm3, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -701,7 +701,7 @@ define i64 @test_v16i64(<16 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm4, %ymm0, %ymm2
 ; AVX2-NEXT:    vxorpd %ymm4, %ymm1, %ymm3
 ; AVX2-NEXT:    vpcmpgtq %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/vector-reduce-umin-widen.ll b/test/CodeGen/X86/vector-reduce-umin-widen.ll
index e5cbd1b73e5e0ae14ce3bdb71b085c414ecd1458..8cdf00c22a64cd3da3b7eff6f6e68d5218ce7ece 100644 (file)
--- a/test/CodeGen/X86/vector-reduce-umin-widen.ll
+++ b/test/CodeGen/X86/vector-reduce-umin-widen.ll
@@ -177,7 +177,7 @@ define i64 @test_v4i64(<4 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpgtq %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -370,7 +370,7 @@ define i64 @test_v8i64(<8 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpgtq %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -700,7 +700,7 @@ define i64 @test_v16i64(<16 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm4, %ymm0, %ymm2
 ; AVX2-NEXT:    vxorpd %ymm4, %ymm1, %ymm3
 ; AVX2-NEXT:    vpcmpgtq %ymm2, %ymm3, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq

diff --git a/test/CodeGen/X86/vector-reduce-umin.ll b/test/CodeGen/X86/vector-reduce-umin.ll
index 258cf4623425c06143683cf4c89e74d80756bceb..75c9b103121f6c30158fa3132c297cb5448edf13 100644 (file)
--- a/test/CodeGen/X86/vector-reduce-umin.ll
+++ b/test/CodeGen/X86/vector-reduce-umin.ll
@@ -177,7 +177,7 @@ define i64 @test_v4i64(<4 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpgtq %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -370,7 +370,7 @@ define i64 @test_v8i64(<8 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm0, %ymm3
 ; AVX2-NEXT:    vxorpd %ymm2, %ymm1, %ymm2
 ; AVX2-NEXT:    vpcmpgtq %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
@@ -700,7 +700,7 @@ define i64 @test_v16i64(<16 x i64> %a0) {
 ; AVX2-NEXT:    vxorpd %ymm4, %ymm0, %ymm2
 ; AVX2-NEXT:    vxorpd %ymm4, %ymm1, %ymm3
 ; AVX2-NEXT:    vpcmpgtq %ymm2, %ymm3, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, %rax
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq