[PATCH] D35218: [AMDGPU] fcanonicalize elimination optimization

[Matt Arsenault via Phabricator via llvm-commits]

arsenm accepted this revision.
arsenm added a comment.
This revision is now accepted and ready to land.

LGTM. Can you file a bug for the broken minnum/maxnum denormal handling



================
Comment at: lib/Target/AMDGPU/SIISelLowering.cpp:4706
+
+    if (IsIEEEMode && isCanonicalized(N0, getSubtarget()))
+      return N0;
----------------
This would read more accurately as IsIEEEMode || isKnownNeverSNan. 


https://reviews.llvm.org/D35218