[PATCH] D18949: [AArch64] Cannonicalize commutative operands based on LSLFast

Chad Rosier via llvm-commits llvm-commits at lists.llvm.org
Tue Apr 19 14:07:45 PDT 2016


mcrosier added a comment.

> ! In http://reviews.llvm.org/D18949#404513, @t.p.northover wrote:

>  I'm instantly suspicious of code behind any kind of CPU check. I'd say enable it generally unless there's good reason not to. It avoids code complication and overly-specific optimizations.


I tend to agree with Tim.  I'd prefer to minimize the use of CPU checks.

Given the limited gains (all appear to be well within noise), I'm inclined to abandon this patch.

Chad


Repository:
  rL LLVM

http://reviews.llvm.org/D18949





More information about the llvm-commits mailing list