[PATCH] D55314: AMDGPU: Turn on the DPP combiner by default

[Connor Abbott via Phabricator via llvm-commits]

cwabbott added a comment.

In D55314#1321326 <https://reviews.llvm.org/D55314#1321326>, @vpykhtin wrote:

> Ok, I'll disable it. I'm not sure about 3rd point: are you sayng the pass doesn't actually perform the optimization or it's fundamentally wrong? Because it implemented to handle "identity" cases for add, mul and min/max.


Ah right, you are handling it. But it seems you're missing and, or, and add. We can't rely on bound_ctrl:0, even if OldOpndValue is zero, so we should always try to fold the operand first if possible.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D55314/new/

https://reviews.llvm.org/D55314