[all-commits] [llvm/llvm-project] 635993: [AMDGPU] Skip unclusterd rescheduling w/o ld/st

Stanislav Mekhanoshin via All-commits all-commits at lists.llvm.org
Fri Feb 26 12:29:25 PST 2021


  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 635993f07bd617818097286eca0bdacb91a9afb5
      https://github.com/llvm/llvm-project/commit/635993f07bd617818097286eca0bdacb91a9afb5
  Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin at amd.com>
  Date:   2021-02-26 (Fri, 26 Feb 2021)

  Changed paths:
    M llvm/lib/Target/AMDGPU/GCNSchedStrategy.cpp
    M llvm/lib/Target/AMDGPU/GCNSchedStrategy.h

  Log Message:
  -----------
  [AMDGPU] Skip unclusterd rescheduling w/o ld/st

We are attempting rescheduling without load store clustering
if occupancy limits were not met with clustering. Skip this
for regions which do not have any loads or stores at all.

In a set of kernels I am experimenting with this improves
scheduling time by ~30%.

Differential Revision: https://reviews.llvm.org/D97342




More information about the All-commits mailing list