[llvm] r294101 - [X86][SSE] Add target shuffle combine buildvec style tests

Simon Pilgrim via llvm-commits llvm-commits at lists.llvm.org
Sat Feb 4 14:17:22 PST 2017


Author: rksimon
Date: Sat Feb  4 16:17:22 2017
New Revision: 294101

URL: http://llvm.org/viewvc/llvm-project?rev=294101&view=rev
Log:
[X86][SSE] Add target shuffle combine buildvec style tests

Extra tests for D29399

Modified:
    llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx2.ll
    llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll

Modified: llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx2.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx2.ll?rev=294101&r1=294100&r2=294101&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx2.ll (original)
+++ llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx2.ll Sat Feb  4 16:17:22 2017
@@ -712,6 +712,35 @@ define <32 x i8> @combine_psrlq_pshufb(<
   ret <32 x i8> %3
 }
 
+define <32 x i8> @combine_unpack_unpack_pshufb(<32 x i8> %a0) {
+; X32-LABEL: combine_unpack_unpack_pshufb:
+; X32:       # BB#0:
+; X32-NEXT:    vpshufd {{.*#+}} ymm1 = ymm0[0,0,0,0,4,4,4,4]
+; X32-NEXT:    vpshufd {{.*#+}} ymm2 = ymm0[1,1,2,3,5,5,6,7]
+; X32-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
+; X32-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[16],ymm2[16],ymm1[17],ymm2[17],ymm1[18],ymm2[18],ymm1[19],ymm2[19],ymm1[20],ymm2[20],ymm1[21],ymm2[21],ymm1[22],ymm2[22],ymm1[23],ymm2[23]
+; X32-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[2],ymm0[2],ymm1[3],ymm0[3],ymm1[4],ymm0[4],ymm1[5],ymm0[5],ymm1[6],ymm0[6],ymm1[7],ymm0[7],ymm1[16],ymm0[16],ymm1[17],ymm0[17],ymm1[18],ymm0[18],ymm1[19],ymm0[19],ymm1[20],ymm0[20],ymm1[21],ymm0[21],ymm1[22],ymm0[22],ymm1[23],ymm0[23]
+; X32-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm2[0],ymm0[0],ymm2[1],ymm0[1],ymm2[2],ymm0[2],ymm2[3],ymm0[3],ymm2[4],ymm0[4],ymm2[5],ymm0[5],ymm2[6],ymm0[6],ymm2[7],ymm0[7],ymm2[16],ymm0[16],ymm2[17],ymm0[17],ymm2[18],ymm0[18],ymm2[19],ymm0[19],ymm2[20],ymm0[20],ymm2[21],ymm0[21],ymm2[22],ymm0[22],ymm2[23],ymm0[23]
+; X32-NEXT:    retl
+;
+; X64-LABEL: combine_unpack_unpack_pshufb:
+; X64:       # BB#0:
+; X64-NEXT:    vpshufd {{.*#+}} ymm1 = ymm0[0,0,0,0,4,4,4,4]
+; X64-NEXT:    vpshufd {{.*#+}} ymm2 = ymm0[1,1,2,3,5,5,6,7]
+; X64-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,2,3,6,7,6,7]
+; X64-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[16],ymm2[16],ymm1[17],ymm2[17],ymm1[18],ymm2[18],ymm1[19],ymm2[19],ymm1[20],ymm2[20],ymm1[21],ymm2[21],ymm1[22],ymm2[22],ymm1[23],ymm2[23]
+; X64-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[2],ymm0[2],ymm1[3],ymm0[3],ymm1[4],ymm0[4],ymm1[5],ymm0[5],ymm1[6],ymm0[6],ymm1[7],ymm0[7],ymm1[16],ymm0[16],ymm1[17],ymm0[17],ymm1[18],ymm0[18],ymm1[19],ymm0[19],ymm1[20],ymm0[20],ymm1[21],ymm0[21],ymm1[22],ymm0[22],ymm1[23],ymm0[23]
+; X64-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm2[0],ymm0[0],ymm2[1],ymm0[1],ymm2[2],ymm0[2],ymm2[3],ymm0[3],ymm2[4],ymm0[4],ymm2[5],ymm0[5],ymm2[6],ymm0[6],ymm2[7],ymm0[7],ymm2[16],ymm0[16],ymm2[17],ymm0[17],ymm2[18],ymm0[18],ymm2[19],ymm0[19],ymm2[20],ymm0[20],ymm2[21],ymm0[21],ymm2[22],ymm0[22],ymm2[23],ymm0[23]
+; X64-NEXT:    retq
+  %1 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 16, i32 17, i32 18, i32 19, i32 16, i32 17, i32 18, i32 19, i32 16, i32 17, i32 18, i32 19>
+  %2 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 20, i32 21, i32 22, i32 23, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+  %3 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 24, i32 25, i32 26, i32 27, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+  %4 = shufflevector <32 x i8> %1, <32 x i8> %2, <32 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+  %5 = shufflevector <32 x i8> %1, <32 x i8> %3, <32 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+  %6 = shufflevector <32 x i8> %4, <32 x i8> %5, <32 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 4, i32 36, i32 5, i32 37, i32 6, i32 38, i32 7, i32 39, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 20, i32 52, i32 21, i32 53, i32 22, i32 54, i32 23, i32 55>
+  ret <32 x i8> %6
+}
+
 define <8 x i32> @constant_fold_permd() {
 ; X32-LABEL: constant_fold_permd:
 ; X32:       # BB#0:

Modified: llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll?rev=294101&r1=294100&r2=294101&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll (original)
+++ llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-ssse3.ll Sat Feb  4 16:17:22 2017
@@ -552,6 +552,46 @@ define <16 x i8> @combine_unpckl_arg1_ps
   ret <16 x i8> %2
 }
 
+define <8 x i16> @shuffle_combine_unpack_insert(<8 x i16> %a0) {
+; SSE-LABEL: shuffle_combine_unpack_insert:
+; SSE:       # BB#0:
+; SSE-NEXT:    pextrw $2, %xmm0, %eax
+; SSE-NEXT:    pextrw $4, %xmm0, %ecx
+; SSE-NEXT:    movdqa %xmm0, %xmm2
+; SSE-NEXT:    pinsrw $4, %eax, %xmm2
+; SSE-NEXT:    movdqa %xmm0, %xmm1
+; SSE-NEXT:    pinsrw $2, %ecx, %xmm1
+; SSE-NEXT:    movdqa %xmm2, %xmm3
+; SSE-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
+; SSE-NEXT:    movdqa %xmm3, %xmm1
+; SSE-NEXT:    punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
+; SSE-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
+; SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
+; SSE-NEXT:    movdqa %xmm1, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: shuffle_combine_unpack_insert:
+; AVX:       # BB#0:
+; AVX-NEXT:    vpextrw $2, %xmm0, %eax
+; AVX-NEXT:    vpextrw $4, %xmm0, %ecx
+; AVX-NEXT:    vpinsrw $4, %eax, %xmm0, %xmm1
+; AVX-NEXT:    vpinsrw $2, %ecx, %xmm0, %xmm2
+; AVX-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; AVX-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; AVX-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
+; AVX-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; AVX-NEXT:    retq
+  %1 = extractelement <8 x i16> %a0, i32 2
+  %2 = extractelement <8 x i16> %a0, i32 4
+  %3 = insertelement <8 x i16> %a0, i16 %1, i32 4
+  %4 = insertelement <8 x i16> %a0, i16 %2, i32 2
+  %5 = shufflevector <8 x i16> %3, <8 x i16> %4, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11>
+  %6 = shufflevector <8 x i16> %5, <8 x i16> %3, <8 x i32> <i32 4, i32 12, i32 5, i32 13, i32 undef, i32 undef, i32 undef, i32 undef>
+  %7 = shufflevector <8 x i16> %5, <8 x i16> %a0, <8 x i32> <i32 4, i32 12, i32 5, i32 13, i32 undef, i32 undef, i32 undef, i32 undef>
+  %8 = shufflevector <8 x i16> %6, <8 x i16> %7, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11>
+  ret <8 x i16> %8
+}
+
 define <16 x i8> @constant_fold_pshufb() {
 ; SSE-LABEL: constant_fold_pshufb:
 ; SSE:       # BB#0:




More information about the llvm-commits mailing list