[PATCH] [x86] Teach the target combine step to aggressively fold pshufd insturcions.

Jim Grosbach grosbach at apple.com
Wed Jun 25 11:11:01 PDT 2014


LGTM

================
Comment at: test/CodeGen/X86/vector-shuffle-combining.ll:12
@@ +11,3 @@
+; CHECK-SSE2-LABEL: @combine_pshufd1
+; CHECK-SSE2:       # BB#0:
+; CHECK-SSE2-NEXT:    retq
----------------
Same -asm-verbose=false comment as on the other patch.

http://reviews.llvm.org/D4292






More information about the llvm-commits mailing list