[all-commits] [llvm/llvm-project] 632956: [mlir][AMDGPU] Explicitly truncate memory addresse...

Krzysztof Drewniak via All-commits all-commits at lists.llvm.org
Thu Aug 4 12:42:48 PDT 2022

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 63295622491a31eaccb6c534ba5caa836beb843f
  Author: Krzysztof Drewniak <Krzysztof.Drewniak at amd.com>
  Date:   2022-08-04 (Thu, 04 Aug 2022)

  Changed paths:
    M mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
    M mlir/test/Conversion/AMDGPUToROCDL/amdgpu-to-rocdl.mlir

  Log Message:
  [mlir][AMDGPU] Explicitly truncate memory addresses in buffer ops

As a percaution, truncate memory addresses passed to kernels to 48 bits,
since bits 48-63 of the buffer descriptor are used for the stride field
and, on gfx10, to control swizzling.

Reviewed By: ThomasRaoux

Differential Revision: https://reviews.llvm.org/D131016

More information about the All-commits mailing list