[x86] Add a bunch of test cases to 256-bit shuffles that exercise

merging 128-bit subvectors and also shuffling all the elements of those
subvectors. Currently we generate pretty bad code for many of these, but
I'm testing a patch that should dramatically improve this in addition to
making the shuffle lowering robust to other changes.

llvm-svn: 222525

3 files changed, 338 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll b/llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll
index 4db0280644b..b3fba4f349b 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-256-v16.ll
@@ -1265,3 +1265,100 @@ define <16 x i16> @shuffle_v16i16_04_04_04_04_uu_uu_uu_uu_08_08_08_uu_uu_12_12_1
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 4, i32 4, i32 4, i32 undef, i32 undef, i32 undef, i32 undef, i32 8, i32 8, i32 8, i32 undef, i32 undef, i32 12, i32 12, i32 12>
   ret <16 x i16> %shuffle
 }
+
+define <16 x i16> @shuffle_v16i16_00_00_00_00_04_04_04_04_16_16_16_16_20_20_20_20(<16 x i16> %a, <16 x i16> %b) {
+; AVX1-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_04_16_16_16_16_20_20_20_20:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_04_16_16_16_16_20_20_20_20:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 4, i32 4, i32 4, i32 16, i32 16, i32 16, i32 16, i32 20, i32 20, i32 20, i32 20>
+  ret <16 x i16> %shuffle
+}
+
+define <16 x i16> @shuffle_v16i16_08_08_08_08_12_12_12_12_16_16_16_16_20_20_20_20(<16 x i16> %a, <16 x i16> %b) {
+; AVX1-LABEL: shuffle_v16i16_08_08_08_08_12_12_12_12_16_16_16_16_20_20_20_20:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v16i16_08_08_08_08_12_12_12_12_16_16_16_16_20_20_20_20:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm1
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm0
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 8, i32 8, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12, i32 16, i32 16, i32 16, i32 16, i32 20, i32 20, i32 20, i32 20>
+  ret <16 x i16> %shuffle
+}
+
+define <16 x i16> @shuffle_v16i16_08_08_08_08_12_12_12_12_24_24_24_24_28_28_28_28(<16 x i16> %a, <16 x i16> %b) {
+; AVX1-LABEL: shuffle_v16i16_08_08_08_08_12_12_12_12_24_24_24_24_28_28_28_28:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v16i16_08_08_08_08_12_12_12_12_24_24_24_24_28_28_28_28:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm0
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm1
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 8, i32 8, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12, i32 24, i32 24, i32 24, i32 24, i32 28, i32 28, i32 28, i32 28>
+  ret <16 x i16> %shuffle
+}
+
+define <16 x i16> @shuffle_v16i16_00_00_00_00_04_04_04_04_24_24_24_24_28_28_28_28(<16 x i16> %a, <16 x i16> %b) {
+; AVX1-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_04_24_24_24_24_28_28_28_28:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_04_24_24_24_24_28_28_28_28:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm1
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 4, i32 4, i32 4, i32 24, i32 24, i32 24, i32 24, i32 28, i32 28, i32 28, i32 28>
+  ret <16 x i16> %shuffle
+}
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-256-v32.ll b/llvm/test/CodeGen/X86/vector-shuffle-256-v32.ll
index 79c906b209a..1c09ba0861d 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-256-v32.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-256-v32.ll
@@ -1560,3 +1560,92 @@ define <32 x i8> @shuffle_v32i8_42_45_12_13_35_35_60_40_17_22_29_44_33_12_48_51_
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 42, i32 45, i32 12, i32 13, i32 35, i32 35, i32 60, i32 40, i32 17, i32 22, i32 29, i32 44, i32 33, i32 12, i32 48, i32 51, i32 20, i32 19, i32 52, i32 19, i32 49, i32 54, i32 37, i32 32, i32 48, i32 42, i32 59, i32 7, i32 36, i32 34, i32 36, i32 39>
   ret <32 x i8> %shuffle
 }
+
+define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_32_32_32_32_32_32_32_32_40_40_40_40_40_40_40_40(<32 x i8> %a, <32 x i8> %b) {
+; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_32_32_32_32_32_32_32_32_40_40_40_40_40_40_40_40:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
+; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_32_32_32_32_32_32_32_32_40_40_40_40_40_40_40_40:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
+; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 40, i32 40, i32 40, i32 40, i32 40, i32 40, i32 40, i32 40>
+  ret <32 x i8> %shuffle
+}
+
+define <32 x i8> @shuffle_v32i8_16_16_16_16_16_16_16_16_24_24_24_24_24_24_24_24_32_32_32_32_32_32_32_32_40_40_40_40_40_40_40_40(<32 x i8> %a, <32 x i8> %b) {
+; AVX1-LABEL: shuffle_v32i8_16_16_16_16_16_16_16_16_24_24_24_24_24_24_24_24_32_32_32_32_32_32_32_32_40_40_40_40_40_40_40_40:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
+; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v32i8_16_16_16_16_16_16_16_16_24_24_24_24_24_24_24_24_32_32_32_32_32_32_32_32_40_40_40_40_40_40_40_40:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
+; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm1
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 40, i32 40, i32 40, i32 40, i32 40, i32 40, i32 40, i32 40>
+  ret <32 x i8> %shuffle
+}
+
+define <32 x i8> @shuffle_v32i8_16_16_16_16_16_16_16_16_24_24_24_24_24_24_24_24_48_48_48_48_48_48_48_48_56_56_56_56_56_56_56_56(<32 x i8> %a, <32 x i8> %b) {
+; AVX1-LABEL: shuffle_v32i8_16_16_16_16_16_16_16_16_24_24_24_24_24_24_24_24_48_48_48_48_48_48_48_48_56_56_56_56_56_56_56_56:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
+; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v32i8_16_16_16_16_16_16_16_16_24_24_24_24_24_24_24_24_48_48_48_48_48_48_48_48_56_56_56_56_56_56_56_56:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm1
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
+; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm0
+; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 48, i32 48, i32 48, i32 48, i32 48, i32 48, i32 48, i32 48, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56>
+  ret <32 x i8> %shuffle
+}
+
+define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_48_48_48_48_48_48_48_48_56_56_56_56_56_56_56_56(<32 x i8> %a, <32 x i8> %b) {
+; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_48_48_48_48_48_48_48_48_56_56_56_56_56_56_56_56:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
+; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_48_48_48_48_48_48_48_48_56_56_56_56_56_56_56_56:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm1
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
+; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
+; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 48, i32 48, i32 48, i32 48, i32 48, i32 48, i32 48, i32 48, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56>
+  ret <32 x i8> %shuffle
+}
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-256-v4.ll b/llvm/test/CodeGen/X86/vector-shuffle-256-v4.ll
index 0bd1bd9d611..18b9970f070 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-256-v4.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-256-v4.ll
@@ -300,6 +300,82 @@ define <4 x double> @shuffle_v4f64_0167(<4 x double> %a, <4 x double> %b) {
   ret <4 x double> %shuffle
 }
 
+define <4 x double> @shuffle_v4f64_1054(<4 x double> %a, <4 x double> %b) {
+; AVX1-LABEL: shuffle_v4f64_1054:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4f64_1054:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,1,1,0]
+; AVX2-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
+; AVX2-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
+  ret <4 x double> %shuffle
+}
+
+define <4 x double> @shuffle_v4f64_3254(<4 x double> %a, <4 x double> %b) {
+; AVX1-LABEL: shuffle_v4f64_3254:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4f64_3254:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,1,1,0]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[3,2,2,3]
+; AVX2-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
+  ret <4 x double> %shuffle
+}
+
+define <4 x double> @shuffle_v4f64_3276(<4 x double> %a, <4 x double> %b) {
+; AVX1-LABEL: shuffle_v4f64_3276:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4f64_3276:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermilpd {{.*#+}} ymm1 = ymm1[0,0,3,2]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[3,2,2,3]
+; AVX2-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
+  ret <4 x double> %shuffle
+}
+
+define <4 x double> @shuffle_v4f64_1076(<4 x double> %a, <4 x double> %b) {
+; AVX1-LABEL: shuffle_v4f64_1076:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4f64_1076:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermilpd {{.*#+}} ymm1 = ymm1[0,0,3,2]
+; AVX2-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[1,0,2,2]
+; AVX2-NEXT:    vblendpd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
+  ret <4 x double> %shuffle
+}
+
 define <4 x i64> @shuffle_v4i64_0000(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-LABEL: shuffle_v4i64_0000:
 ; AVX1:       # BB#0:
@@ -608,6 +684,82 @@ define <4 x i64> @shuffle_v4i64_1251(<4 x i64> %a, <4 x i64> %b) {
   ret <4 x i64> %shuffle
 }
 
+define <4 x i64> @shuffle_v4i64_1054(<4 x i64> %a, <4 x i64> %b) {
+; AVX1-LABEL: shuffle_v4i64_1054:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i64_1054:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,1,1,0]
+; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 5, i32 4>
+  ret <4 x i64> %shuffle
+}
+
+define <4 x i64> @shuffle_v4i64_3254(<4 x i64> %a, <4 x i64> %b) {
+; AVX1-LABEL: shuffle_v4i64_3254:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i64_3254:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,1,1,0]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[3,2,2,3]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 5, i32 4>
+  ret <4 x i64> %shuffle
+}
+
+define <4 x i64> @shuffle_v4i64_3276(<4 x i64> %a, <4 x i64> %b) {
+; AVX1-LABEL: shuffle_v4i64_3276:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i64_3276:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[2,3,0,1,6,7,4,5]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[3,2,2,3]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 3, i32 2, i32 7, i32 6>
+  ret <4 x i64> %shuffle
+}
+
+define <4 x i64> @shuffle_v4i64_1076(<4 x i64> %a, <4 x i64> %b) {
+; AVX1-LABEL: shuffle_v4i64_1076:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vpermilpd {{.*#+}} xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i64_1076:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm1[2,3,0,1,6,7,4,5]
+; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; AVX2-NEXT:    retq
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 1, i32 0, i32 7, i32 6>
+  ret <4 x i64> %shuffle
+}
+
 define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-LABEL: stress_test1:
 ; AVX1:       # BB#0: