[PATCH] D34559: [X86][DAG] Switch X86 Target to post-legalized store merge

Simon Pilgrim via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Sat Jun 24 06:16:37 PDT 2017


RKSimon added inline comments.


================
Comment at: test/CodeGen/X86/oddshuffles.ll:86
 ; SSE42-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
-; SSE42-NEXT:    pextrd $2, %xmm0, 8(%rdi)
-; SSE42-NEXT:    movq %xmm1, (%rdi)
+; SSE42-NEXT:    movdqa %xmm1, (%rdi)
 ; SSE42-NEXT:    retq
----------------
This is scaring me - what is the reasoning behind permitting the store of a <4 x i32> and not just the <3 x i32> ?


https://reviews.llvm.org/D34559





More information about the llvm-commits mailing list