[PATCH] D23919: [X86] Loosen memory folding requirements for cvtdq2pd and cvtps2pd instructions

Simon Pilgrim via llvm-commits llvm-commits at lists.llvm.org
Mon Aug 29 09:04:39 PDT 2016

RKSimon added a reviewer: RKSimon.

Comment at: test/CodeGen/X86/peephole-cvt-sse.ll:2
@@ +1,3 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=x86_64-pc-linux -mcpu=slm < %s | FileCheck %s --check-prefix=X86-64
+; RUN: llc -mtriple=i386-pc-linux -mcpu=slm < %s | FileCheck %s --check-prefix=I386
aturetsk wrote:
> Both are done.
Thanks, its also better to use -mattr=+sse4.2 (or similar) instead of a cpu target unless you are specifically testings for that cpu.


More information about the llvm-commits mailing list