[PATCH] D33802: [AMDGPU] Preserve operand order in SIFoldOperands

[Stanislav Mekhanoshin via Phabricator via llvm-commits]

rampitec added a comment.

In https://reviews.llvm.org/D33802#770737, @arsenm wrote:

> It would make sense to compute even if nothing is folded, since having constants in src0 as the canonical order would make sense, although I think it blindly commutes still as it is


It now blindly commutes everything.


Repository:
  rL LLVM

https://reviews.llvm.org/D33802