[PATCH] D40633: [PCG] Poor shuffle lane tracking (PR35454 )

Sanjay Patel via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Fri Apr 3 10:15:10 PDT 2020


spatel added a comment.

In D40633#1958062 <https://reviews.llvm.org/D40633#1958062>, @RKSimon wrote:

> @spatel Maybe ensure we have all the test coverage from the tests that Konstantin added here?


Tests adapted from this patch and added to "PhaseOrdering":
rG389704cc601 <https://reviews.llvm.org/rG389704cc601bef026a4000ecf6ed81eed2a7c326>

I added test comments about what we still can do to improve things. Also note that the fold in -vector-combine relies on the cost model, so we don't get the simplifications for a base SSE2 compile (I assumed from the asm shown in https://bugs.llvm.org/show_bug.cgi?id=35454 that we care about an AVX or later target).


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D40633/new/

https://reviews.llvm.org/D40633





More information about the llvm-commits mailing list