[PATCH] D111754: AMDGPU: Fixes for 'LOD bias' operand in ISelDAG path and GobalISel path when A16-bit is 'ON'

[Sebastian Neubauer via Phabricator via llvm-commits]

sebastian-ne added a comment.

This also needs fixing in the combiner (which was introduced in D85887 <https://reviews.llvm.org/D85887>). I.e. checking that the bias can be losslessly converted to f16 and converting it when converting the address.
(I guess this doesn’t need to be part of this change.)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D111754/new/

https://reviews.llvm.org/D111754