[PATCH] D69654: AMDGPU: Switch backend default max workgroup size to 1024

[Matt Arsenault via Phabricator via llvm-commits]

arsenm created this revision.
arsenm added reviewers: kzhuravl, rampitec.
Herald added subscribers: hiraditya, t-tye, tpr, dstuttard, yaxunl, nhaehnle, wdng, jvesely.
Herald added a project: LLVM.

Previously this would default to 256, not the maximum supported size
of 1024. Using a maximum lower than the hardware maximum requires
language runtimes to enforce this limit for correctness, which no
language has correctly done. Switch the default to the conservatively
correct maximum, and force frontends to opt-in to the more optimal 256
default maximum.

      

I don't really understand why the changes in occupancy-levels.ll
increased the computed occupancy, which I expected to decrease. I'm
not sure if these tests should be forcing the old maximum.


https://reviews.llvm.org/D69654

Files:
  llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
  llvm/test/CodeGen/AMDGPU/amdgpu.private-memory.ll
  llvm/test/CodeGen/AMDGPU/array-ptr-calc-i32.ll
  llvm/test/CodeGen/AMDGPU/hsa-metadata-kernel-code-props-v3.ll
  llvm/test/CodeGen/AMDGPU/hsa-metadata-kernel-code-props.ll
  llvm/test/CodeGen/AMDGPU/lower-range-metadata-intrinsic-call.ll
  llvm/test/CodeGen/AMDGPU/occupancy-levels.ll
  llvm/test/CodeGen/AMDGPU/private-memory-r600.ll
  llvm/test/CodeGen/AMDGPU/promote-alloca-addrspacecast.ll
  llvm/test/CodeGen/AMDGPU/promote-alloca-to-lds-icmp.ll
  llvm/test/CodeGen/AMDGPU/promote-alloca-to-lds-phi.ll
  llvm/test/CodeGen/AMDGPU/promote-alloca-to-lds-select.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D69654.227222.patch
Type: text/x-patch
Size: 9772 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20191031/f5223e75/attachment.bin>