[all-commits] [llvm/llvm-project] 288085: [mlir][ROCDL] Remove unneeded bf16 expansion in Lo...

[Krzysztof Drewniak via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 28808596040734e1b373af722dfaf0158174b534
      https://github.com/llvm/llvm-project/commit/28808596040734e1b373af722dfaf0158174b534
  Author: Krzysztof Drewniak <Krzysztof.Drewniak at amd.com>
  Date:   2025-05-12 (Mon, 12 May 2025)

  Changed paths:
    M mlir/lib/Conversion/GPUToROCDL/LowerGpuOpsToROCDLOps.cpp

  Log Message:
  -----------
  [mlir][ROCDL] Remove unneeded bf16 expansion in LowerGPUToROCDL (#139603)

The umbrella pass fol lowering GPU ops to ROCDL (aka  lowering to LLVM
+ the AMDGPU-specific setup) would call the arith patterns that manually
implemented extf and truncf on bfloat because the LLVM AMDGPU backend
used to not suppport those operaitons.

Since the backend does now support these operations and has for quite
some time, remove these patterns from the default lowering flow.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications