[X86][AVX2] Improve lowerShuffleAsRepeatedMaskAndLanePermute permutation of 64-bit sub-lanes

As discussed on PR28136, lowerShuffleAsRepeatedMaskAndLanePermute was attempting to match repeated masks at the 128-bit level and then permute the resultant lanes at the 128-bit (AVX1) or 64-bit (AVX2) sub-lane level.

This change allows us to create the repeated masks at the sub-lane level (and then concat them together to create a 128-bit repeated mask) and then select which sub-lane to permute. This has no effect on the AVX1 codegen.

Fixes PR28136.

llvm-svn: 275543

1 files changed, 34 insertions, 37 deletions

diff --git a/llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll b/llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll
index cede0a95bb9..4aab5cd1700 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-256-v8.ll
@@ -29,8 +29,8 @@ define <8 x float> @shuffle_v8f32_00000010(<8 x float> %a, <8 x float> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8f32_00000010:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [0,0,0,0,0,0,1,0]
-; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,1,0]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,0,1]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 0>
   ret <8 x float> %shuffle
@@ -46,8 +46,8 @@ define <8 x float> @shuffle_v8f32_00000200(<8 x float> %a, <8 x float> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8f32_00000200:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [0,0,0,0,0,2,0,0]
-; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,0,2]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,1,0]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 2, i32 0, i32 0>
   ret <8 x float> %shuffle
@@ -63,8 +63,8 @@ define <8 x float> @shuffle_v8f32_00003000(<8 x float> %a, <8 x float> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8f32_00003000:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [0,0,0,0,3,0,0,0]
-; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,3,0]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,1,0]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 3, i32 0, i32 0, i32 0>
   ret <8 x float> %shuffle
@@ -176,8 +176,8 @@ define <8 x float> @shuffle_v8f32_00001111(<8 x float> %a, <8 x float> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8f32_00001111:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [0,0,0,0,1,1,1,1]
-; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,1,1]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,1,1]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1>
   ret <8 x float> %shuffle
@@ -296,10 +296,10 @@ define <8 x float> @shuffle_v8f32_08991abb(<8 x float> %a, <8 x float> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8f32_08991abb:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <0,u,u,u,1,u,u,u>
-; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <u,0,1,1,u,2,3,3>
 ; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,1,1,3]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 8, i32 9, i32 9, i32 1, i32 10, i32 11, i32 11>
@@ -335,8 +335,8 @@ define <8 x float> @shuffle_v8f32_09ab1def(<8 x float> %a, <8 x float> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8f32_09ab1def:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <0,u,u,u,1,u,u,u>
-; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,1,1,3]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3]
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 1, i32 13, i32 14, i32 15>
@@ -644,10 +644,10 @@ define <8 x float> @shuffle_v8f32_c348cda0(<8 x float> %a, <8 x float> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8f32_c348cda0:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,3,0,0,4,7,4,4]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,1]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[0,1,2,0,4,5,6,4]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[2,1,2,1]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <4,u,u,0,4,5,2,u>
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[0,3,2,0,4,7,6,4]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,1]
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2],ymm1[3,4,5,6],ymm0[7]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 12, i32 3, i32 4, i32 8, i32 12, i32 13, i32 10, i32 0>
@@ -668,10 +668,10 @@ define <8 x float> @shuffle_v8f32_f511235a(<8 x float> %a, <8 x float> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8f32_f511235a:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = <u,5,1,1,2,3,5,u>
-; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[3,1,2,2,7,5,6,6]
-; AVX2-NEXT:    vperm2f128 {{.*#+}} ymm1 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[3,2,2,3,7,6,6,7]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[2,1,2,0]
+; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[1,1,2,3,5,5,6,7]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,0,1,2]
 ; AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0],ymm0[1,2,3,4,5,6],ymm1[7]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 15, i32 5, i32 1, i32 1, i32 2, i32 3, i32 5, i32 10>
@@ -767,11 +767,8 @@ define <8 x float> @PR21138(<8 x float> %truc, <8 x float> %tchose) {
 ;
 ; AVX2-LABEL: PR21138:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm1[1,3,1,3,5,7,5,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,1,0,3]
-; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[1,3,1,3,5,7,5,7]
-; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,3,2,3]
-; AVX2-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
+; AVX2-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[1,3],ymm1[1,3],ymm0[5,7],ymm1[5,7]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,2,1,3]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %truc, <8 x float> %tchose, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
   ret <8 x float> %shuffle
@@ -914,8 +911,8 @@ define <8 x i32> @shuffle_v8i32_00000010(<8 x i32> %a, <8 x i32> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8i32_00000010:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,0,0,0,0,0,1,0]
-; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,0]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,0,0,1]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 0>
   ret <8 x i32> %shuffle
@@ -931,8 +928,8 @@ define <8 x i32> @shuffle_v8i32_00000200(<8 x i32> %a, <8 x i32> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8i32_00000200:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,0,0,0,0,2,0,0]
-; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,2]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,0,1,0]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 2, i32 0, i32 0>
   ret <8 x i32> %shuffle
@@ -948,8 +945,8 @@ define <8 x i32> @shuffle_v8i32_00003000(<8 x i32> %a, <8 x i32> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8i32_00003000:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,0,0,0,3,0,0,0]
-; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,3,0]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,0,1,0]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 3, i32 0, i32 0, i32 0>
   ret <8 x i32> %shuffle
@@ -1066,8 +1063,8 @@ define <8 x i32> @shuffle_v8i32_00001111(<8 x i32> %a, <8 x i32> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8i32_00001111:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,0,0,0,1,1,1,1]
-; AVX2-NEXT:    vpermd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,0,1,1]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1>
   ret <8 x i32> %shuffle
@@ -1226,10 +1223,10 @@ define <8 x i32> @shuffle_v8i32_08991abb(<8 x i32> %a, <8 x i32> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8i32_08991abb:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,u,u,u,1,u,u,u>
-; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <u,0,1,1,u,2,3,3>
 ; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 8, i32 9, i32 9, i32 1, i32 10, i32 11, i32 11>
@@ -1264,8 +1261,8 @@ define <8 x i32> @shuffle_v8i32_09ab1def(<8 x i32> %a, <8 x i32> %b) {
 ;
 ; AVX2-LABEL: shuffle_v8i32_09ab1def:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,u,u,u,1,u,u,u>
-; AVX2-NEXT:    vpermd %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,1,3]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 1, i32 13, i32 14, i32 15>