[PATCH] D37616: [X86] PR34149 Suboptimal codegen for fast minnum and maxnum.

[Alina Sbirlea via Phabricator via llvm-commits]

asbirlea added a comment.

FWIW, we've seen a similar case of min/maxnum being generated in the IR and leading to suboptimal codegen. 
The solution was replacing the IR generated with fcmp+select (coupled with https://reviews.llvm.org/D27846). Seems like the right approach to avoid adding complexity to the backend.


https://reviews.llvm.org/D37616