[PATCH] D102334: [AMDGPU] Sort LDS globals based on thier size and alignment.

[Mahesha S via Phabricator via llvm-commits]

hsmhsm created this revision.
hsmhsm added reviewers: rampitec, arsenm, JonChesterfield, t-tye, b-sumner, foad.
Herald added subscribers: kerbowa, hiraditya, tpr, dstuttard, yaxunl, mgorny, nhaehnle, jvesely, kzhuravl.
hsmhsm requested review of this revision.
Herald added subscribers: llvm-commits, wdng.
Herald added a project: LLVM.

In order to increase the probability of aligned accessing of LDS memory
operations, sort LDS globals based on thier size and alignment, and then
allocate memory in that sorted order.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D102334

Files:
  llvm/lib/Target/AMDGPU/AMDGPU.h
  llvm/lib/Target/AMDGPU/AMDGPUMachineFunction.cpp
  llvm/lib/Target/AMDGPU/AMDGPUSortLDSGlobals.cpp
  llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
  llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.h
  llvm/lib/Target/AMDGPU/CMakeLists.txt
  llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.cpp
  llvm/lib/Target/AMDGPU/Utils/AMDGPULDSUtils.h

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D102334.344850.patch
Type: text/x-patch
Size: 12087 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210512/b883a017/attachment.bin>