[PATCH] D50328: [X86][SSE] Combine (some) target shuffles with multiple uses

[David Greene via Phabricator via llvm-commits]

greened added inline comments.


================
Comment at: lib/Target/X86/X86ISelLowering.cpp:39640
+    return RHS;
+
   const TargetLowering &TLI = DAG.getTargetLoweringInfo();
----------------
This would be best done as a separate change.


================
Comment at: test/CodeGen/X86/2012-01-12-extract-sv.ll:12
+; CHECK-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0],xmm2[1,2,3]
+; CHECK-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,0,0,0]
 ; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
----------------
Can we make this test less brittle by using FileCheck variables?  This goes for pretty much every test in this patch.


================
Comment at: test/CodeGen/X86/avx512-intrinsics-fast-isel.ll:6498
+; X86-NEXT:    vpmuludq %xmm3, %xmm1, %xmm3
+; X86-NEXT:    vpaddq %xmm2, %xmm3, %xmm2
 ; X86-NEXT:    vpsllq $32, %xmm2, %xmm2
----------------
Make this less brittle with FileCheck variables.


================
Comment at: test/CodeGen/X86/avx512-intrinsics-fast-isel.ll:6740
+; X64-NEXT:    vpmuludq %xmm3, %xmm1, %xmm3
+; X64-NEXT:    vpaddq %xmm2, %xmm3, %xmm2
 ; X64-NEXT:    vpsllq $32, %xmm2, %xmm2
----------------
Make this less brittle with FileCheck variables.


Repository:
  rL LLVM

https://reviews.llvm.org/D50328