diff options
| author | Simon Pilgrim <llvm-dev@redking.me.uk> | 2018-02-16 14:57:25 +0000 |
|---|---|---|
| committer | Simon Pilgrim <llvm-dev@redking.me.uk> | 2018-02-16 14:57:25 +0000 |
| commit | 4e2f757dc100a2bf75b8dc9d6302ebd453b3cd2c (patch) | |
| tree | ff6b7eb3ef360a1bc57e49d9cb589a9e4ab4fc52 /llvm/test/CodeGen/X86/oddshuffles.ll | |
| parent | 6cf41b028d5deaee192d8e9195af2b9a80dea3ed (diff) | |
| download | bcm5719-llvm-4e2f757dc100a2bf75b8dc9d6302ebd453b3cd2c.tar.gz bcm5719-llvm-4e2f757dc100a2bf75b8dc9d6302ebd453b3cd2c.zip | |
[X86][SSE] Allow float domain crossing if we are merging 2 or more shuffles and the root started as a float domain shuffle
llvm-svn: 325349
Diffstat (limited to 'llvm/test/CodeGen/X86/oddshuffles.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/oddshuffles.ll | 37 |
1 files changed, 17 insertions, 20 deletions
diff --git a/llvm/test/CodeGen/X86/oddshuffles.ll b/llvm/test/CodeGen/X86/oddshuffles.ll index 1fd4e0b0214..05cbf58d794 100644 --- a/llvm/test/CodeGen/X86/oddshuffles.ll +++ b/llvm/test/CodeGen/X86/oddshuffles.ll @@ -614,18 +614,17 @@ define void @v12i32(<8 x i32> %a, <8 x i32> %b, <12 x i32>* %p) nounwind { ; SSE2-NEXT: movdqa %xmm0, %xmm3 ; SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,1,2,2] -; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm2[0,1,0,1] -; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[2,0],xmm3[3,0] +; SSE2-NEXT: movaps %xmm2, %xmm4 +; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,0],xmm3[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,1],xmm4[0,2] -; SSE2-NEXT: movdqa %xmm2, %xmm4 +; SSE2-NEXT: movaps %xmm2, %xmm4 ; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[1,0],xmm1[1,0] ; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[2,0],xmm1[2,2] -; SSE2-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm0[2],xmm2[3],xmm0[3] +; SSE2-NEXT: unpckhps {{.*#+}} xmm2 = xmm2[2],xmm0[2],xmm2[3],xmm0[3] ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,0],xmm4[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,1],xmm0[0,2] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[0,3,2,2] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,2,3,3] -; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0] +; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[3,2],xmm0[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2] ; SSE2-NEXT: movaps %xmm0, 32(%rdi) ; SSE2-NEXT: movaps %xmm4, 16(%rdi) @@ -1562,39 +1561,37 @@ define void @interleave_24i32_in(<24 x i32>* %p, <8 x i32>* %q1, <8 x i32>* %q2, ; SSE2-NEXT: movdqu 16(%rsi), %xmm2 ; SSE2-NEXT: movdqu (%rdx), %xmm6 ; SSE2-NEXT: movdqu 16(%rdx), %xmm1 -; SSE2-NEXT: movdqu (%rcx), %xmm7 -; SSE2-NEXT: movdqu 16(%rcx), %xmm4 +; SSE2-NEXT: movups (%rcx), %xmm7 +; SSE2-NEXT: movups 16(%rcx), %xmm4 ; SSE2-NEXT: movdqa %xmm5, %xmm0 ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,2,2] -; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm7[0,1,0,1] -; SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[2,0],xmm0[3,0] +; SSE2-NEXT: movaps %xmm7, %xmm3 +; SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,0],xmm0[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm3[0,2] -; SSE2-NEXT: movdqa %xmm7, %xmm3 +; SSE2-NEXT: movaps %xmm7, %xmm3 ; SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[1,0],xmm6[1,0] ; SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[2,0],xmm6[2,2] -; SSE2-NEXT: punpckhdq {{.*#+}} xmm7 = xmm7[2],xmm5[2],xmm7[3],xmm5[3] +; SSE2-NEXT: unpckhps {{.*#+}} xmm7 = xmm7[2],xmm5[2],xmm7[3],xmm5[3] ; SSE2-NEXT: shufps {{.*#+}} xmm5 = xmm5[2,0],xmm3[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,1],xmm5[0,2] ; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm7[0,3,2,2] -; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm6[2,2,3,3] -; SSE2-NEXT: shufps {{.*#+}} xmm6 = xmm6[2,0],xmm5[3,0] +; SSE2-NEXT: shufps {{.*#+}} xmm6 = xmm6[3,2],xmm5[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm5 = xmm5[0,1],xmm6[0,2] ; SSE2-NEXT: movdqa %xmm2, %xmm6 ; SSE2-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm1[0],xmm6[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm6[0,1,2,2] -; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm4[0,1,0,1] -; SSE2-NEXT: shufps {{.*#+}} xmm7 = xmm7[2,0],xmm6[3,0] +; SSE2-NEXT: movaps %xmm4, %xmm7 +; SSE2-NEXT: shufps {{.*#+}} xmm7 = xmm7[0,0],xmm6[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm6 = xmm6[0,1],xmm7[0,2] -; SSE2-NEXT: movdqa %xmm4, %xmm7 +; SSE2-NEXT: movaps %xmm4, %xmm7 ; SSE2-NEXT: shufps {{.*#+}} xmm7 = xmm7[1,0],xmm1[1,0] ; SSE2-NEXT: shufps {{.*#+}} xmm7 = xmm7[2,0],xmm1[2,2] -; SSE2-NEXT: punpckhdq {{.*#+}} xmm4 = xmm4[2],xmm2[2],xmm4[3],xmm2[3] +; SSE2-NEXT: unpckhps {{.*#+}} xmm4 = xmm4[2],xmm2[2],xmm4[3],xmm2[3] ; SSE2-NEXT: shufps {{.*#+}} xmm2 = xmm2[2,0],xmm7[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm7 = xmm7[0,1],xmm2[0,2] ; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm4[0,3,2,2] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,2,3,3] -; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[2,0],xmm2[3,0] +; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[3,2],xmm2[3,0] ; SSE2-NEXT: shufps {{.*#+}} xmm2 = xmm2[0,1],xmm1[0,2] ; SSE2-NEXT: movups %xmm2, 80(%rdi) ; SSE2-NEXT: movups %xmm7, 64(%rdi) |
