[PATCH] D19087: [x86, ppc] prefer comparisons against zero for and+cmp sequences

Sanjay Patel via llvm-commits llvm-commits at lists.llvm.org
Thu Apr 14 12:40:03 PDT 2016


spatel added a comment.

In http://reviews.llvm.org/D19087#401557, @hfinkel wrote:

> Not sure why PPC would take more work than x86 (both backends currently use the MachineCombiner for other purposes). Regardless, I'd not jump here unless necessary.


In the test cases, PPC is branching, so I figure there'd be more complexity undoing that than the simple instruction replacement/deletion that x86 needs. But I agree, I don't want to sink that far down if I don't have to.


http://reviews.llvm.org/D19087





More information about the llvm-commits mailing list