[X86][SSE] Add shuffle combining tests with PACKSS/PACKUS

llvm-svn: 314628

2 files changed, 156 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll b/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll
index 671d191b5dd..0e3c4e402a8 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll
@@ -804,6 +804,96 @@ define <32 x i8> @combine_unpack_unpack_pshufb(<32 x i8> %a0) {
   ret <32 x i8> %6
 }
 
+define <16 x i16> @shuffle_combine_packssdw_pshufb(<8 x i32> %a0) {
+; X32-LABEL: shuffle_combine_packssdw_pshufb:
+; X32:       # BB#0:
+; X32-NEXT:    vpsrad $31, %ymm0, %ymm0
+; X32-NEXT:    vpackssdw %ymm0, %ymm0, %ymm0
+; X32-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[6,7,4,5,2,3,0,1,6,7,4,5,2,3,0,1,16,17,18,19,20,21,22,23,22,23,20,21,18,19,16,17]
+; X32-NEXT:    retl
+;
+; X64-LABEL: shuffle_combine_packssdw_pshufb:
+; X64:       # BB#0:
+; X64-NEXT:    vpsrad $31, %ymm0, %ymm0
+; X64-NEXT:    vpackssdw %ymm0, %ymm0, %ymm0
+; X64-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[6,7,4,5,2,3,0,1,6,7,4,5,2,3,0,1,16,17,18,19,20,21,22,23,22,23,20,21,18,19,16,17]
+; X64-NEXT:    retq
+  %1 = ashr <8 x i32> %a0, <i32 31, i32 31, i32 31, i32 31, i32 31, i32 31, i32 31, i32 31>
+  %2 = tail call <16 x i16> @llvm.x86.avx2.packssdw(<8 x i32> %1, <8 x i32> %1)
+  %3 = shufflevector <16 x i16> %2, <16 x i16> undef, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0, i32 8, i32 9, i32 10, i32 11, i32 11, i32 10, i32 9, i32 8>
+  ret <16 x i16> %3
+}
+declare <16 x i16> @llvm.x86.avx2.packssdw(<8 x i32>, <8 x i32>) nounwind readnone
+
+define <32 x i8> @shuffle_combine_packsswb_pshufb(<16 x i16> %a0, <16 x i16> %a1) {
+; X32-LABEL: shuffle_combine_packsswb_pshufb:
+; X32:       # BB#0:
+; X32-NEXT:    vpsraw $15, %ymm0, %ymm0
+; X32-NEXT:    vpsraw $15, %ymm1, %ymm1
+; X32-NEXT:    vpacksswb %ymm1, %ymm0, %ymm0
+; X32-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[7,6,5,4,3,2,1,0,7,6,5,4,3,2,1,0,23,22,21,20,19,18,17,16,23,22,21,20,19,18,17,16]
+; X32-NEXT:    retl
+;
+; X64-LABEL: shuffle_combine_packsswb_pshufb:
+; X64:       # BB#0:
+; X64-NEXT:    vpsraw $15, %ymm0, %ymm0
+; X64-NEXT:    vpsraw $15, %ymm1, %ymm1
+; X64-NEXT:    vpacksswb %ymm1, %ymm0, %ymm0
+; X64-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[7,6,5,4,3,2,1,0,7,6,5,4,3,2,1,0,23,22,21,20,19,18,17,16,23,22,21,20,19,18,17,16]
+; X64-NEXT:    retq
+  %1 = ashr <16 x i16> %a0, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15>
+  %2 = ashr <16 x i16> %a1, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15>
+  %3 = tail call <32 x i8> @llvm.x86.avx2.packsswb(<16 x i16> %1, <16 x i16> %2)
+  %4 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %3, <32 x i8> <i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+  ret <32 x i8> %4
+}
+declare <32 x i8> @llvm.x86.avx2.packsswb(<16 x i16>, <16 x i16>) nounwind readnone
+
+define <16 x i16> @shuffle_combine_packusdw_pshufb(<8 x i32> %a0, <8 x i32> %a1) {
+; X32-LABEL: shuffle_combine_packusdw_pshufb:
+; X32:       # BB#0:
+; X32-NEXT:    vpand {{\.LCPI.*}}, %ymm0, %ymm0
+; X32-NEXT:    vpackusdw %ymm0, %ymm0, %ymm0
+; X32-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[6,7,4,5,2,3,0,1,6,7,4,5,2,3,0,1,16,17,18,19,20,21,22,23,22,23,20,21,18,19,16,17]
+; X32-NEXT:    retl
+;
+; X64-LABEL: shuffle_combine_packusdw_pshufb:
+; X64:       # BB#0:
+; X64-NEXT:    vpand {{.*}}(%rip), %ymm0, %ymm0
+; X64-NEXT:    vpackusdw %ymm0, %ymm0, %ymm0
+; X64-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[6,7,4,5,2,3,0,1,6,7,4,5,2,3,0,1,16,17,18,19,20,21,22,23,22,23,20,21,18,19,16,17]
+; X64-NEXT:    retq
+  %1 = and <8 x i32> %a0, <i32 255, i32 65535, i32 255, i32 65535, i32 255, i32 255, i32 255, i32 65535>
+  %2 = tail call <16 x i16> @llvm.x86.avx2.packusdw(<8 x i32> %1, <8 x i32> %1)
+  %3 = shufflevector <16 x i16> %2, <16 x i16> undef, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0, i32 8, i32 9, i32 10, i32 11, i32 11, i32 10, i32 9, i32 8>
+  ret <16 x i16> %3
+}
+declare <16 x i16> @llvm.x86.avx2.packusdw(<8 x i32>, <8 x i32>) nounwind readnone
+
+define <32 x i8> @shuffle_combine_packuswb_pshufb(<16 x i16> %a0, <16 x i16> %a1) {
+; X32-LABEL: shuffle_combine_packuswb_pshufb:
+; X32:       # BB#0:
+; X32-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; X32-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; X32-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
+; X32-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[7,6,5,4,3,2,1,0,7,6,5,4,3,2,1,0,23,22,21,20,19,18,17,16,23,22,21,20,19,18,17,16]
+; X32-NEXT:    retl
+;
+; X64-LABEL: shuffle_combine_packuswb_pshufb:
+; X64:       # BB#0:
+; X64-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; X64-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; X64-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
+; X64-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[7,6,5,4,3,2,1,0,7,6,5,4,3,2,1,0,23,22,21,20,19,18,17,16,23,22,21,20,19,18,17,16]
+; X64-NEXT:    retq
+  %1 = lshr <16 x i16> %a0, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8>
+  %2 = lshr <16 x i16> %a1, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8>
+  %3 = tail call <32 x i8> @llvm.x86.avx2.packuswb(<16 x i16> %1, <16 x i16> %2)
+  %4 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %3, <32 x i8> <i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+  ret <32 x i8> %4
+}
+declare <32 x i8> @llvm.x86.avx2.packuswb(<16 x i16>, <16 x i16>) nounwind readnone
+
 define <16 x i8> @combine_pshufb_insertion_as_broadcast_v2i64(i64 %a0) {
 ; X32-LABEL: combine_pshufb_insertion_as_broadcast_v2i64:
 ; X32:       # BB#0:
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll b/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
index a11cac0b9bd..6a88bf010e7 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
@@ -640,6 +640,72 @@ define <8 x i16> @shuffle_combine_unpack_insert(<8 x i16> %a0) {
   ret <8 x i16> %8
 }
 
+define <16 x i8> @shuffle_combine_packssdw_pshufb(<4 x i32> %a0) {
+; SSE-LABEL: shuffle_combine_packssdw_pshufb:
+; SSE:       # BB#0:
+; SSE-NEXT:    psrad $31, %xmm0
+; SSE-NEXT:    packssdw %xmm0, %xmm0
+; SSE-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[7,6,5,4,3,2,1,0,15,14,13,12,11,10,9,8]
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: shuffle_combine_packssdw_pshufb:
+; AVX:       # BB#0:
+; AVX-NEXT:    vpsrad $31, %xmm0, %xmm0
+; AVX-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[7,6,5,4,3,2,1,0,15,14,13,12,11,10,9,8]
+; AVX-NEXT:    retq
+  %1 = ashr <4 x i32> %a0, <i32 31, i32 31, i32 31, i32 31>
+  %2 = tail call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %1, <4 x i32> %1)
+  %3 = bitcast <8 x i16> %2 to <16 x i8>
+  %4 = tail call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %3, <16 x i8> <i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 15, i8 14, i8 13, i8 12, i8 11, i8 10, i8 9, i8 8>)
+  ret <16 x i8> %4
+}
+declare <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32>, <4 x i32>) nounwind readnone
+
+define <16 x i8> @shuffle_combine_packsswb_pshufb(<8 x i16> %a0, <8 x i16> %a1) {
+; SSE-LABEL: shuffle_combine_packsswb_pshufb:
+; SSE:       # BB#0:
+; SSE-NEXT:    psraw $15, %xmm0
+; SSE-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[14,12,10,8,6,4,2,0,14,12,10,8,6,4,2,0]
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: shuffle_combine_packsswb_pshufb:
+; AVX:       # BB#0:
+; AVX-NEXT:    vpsraw $15, %xmm0, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[14,12,10,8,6,4,2,0,14,12,10,8,6,4,2,0]
+; AVX-NEXT:    retq
+  %1 = ashr <8 x i16> %a0, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15>
+  %2 = ashr <8 x i16> %a1, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15>
+  %3 = tail call <16 x i8> @llvm.x86.sse2.packsswb.128(<8 x i16> %1, <8 x i16> %2)
+  %4 = tail call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %3, <16 x i8> <i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+  ret <16 x i8> %4
+}
+declare <16 x i8> @llvm.x86.sse2.packsswb.128(<8 x i16>, <8 x i16>) nounwind readnone
+
+define <16 x i8> @shuffle_combine_packuswb_pshufb(<8 x i16> %a0, <8 x i16> %a1) {
+; SSE-LABEL: shuffle_combine_packuswb_pshufb:
+; SSE:       # BB#0:
+; SSE-NEXT:    psrlw $8, %xmm0
+; SSE-NEXT:    psrlw $8, %xmm1
+; SSE-NEXT:    packuswb %xmm1, %xmm0
+; SSE-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[7,6,5,4,3,2,1,0,7,6,5,4,3,2,1,0]
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: shuffle_combine_packuswb_pshufb:
+; AVX:       # BB#0:
+; AVX-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[7,6,5,4,3,2,1,0,7,6,5,4,3,2,1,0]
+; AVX-NEXT:    retq
+  %1 = lshr <8 x i16> %a0, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8>
+  %2 = lshr <8 x i16> %a1, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8>
+  %3 = tail call <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16> %1, <8 x i16> %2)
+  %4 = tail call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %3, <16 x i8> <i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+  ret <16 x i8> %4
+}
+declare <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16>, <8 x i16>) nounwind readnone
+
 define <16 x i8> @constant_fold_pshufb() {
 ; SSE-LABEL: constant_fold_pshufb:
 ; SSE:       # BB#0: