[PATCH] D18162: AMDGPU: Add SIWholeQuadMode pass

[Nicolai Hähnle via llvm-commits]

nhaehnle updated this revision to Diff 50671.
nhaehnle added a comment.

In the fast case where we have WQM but don't need to switch back and forth,
also mark COPYs as implicitly using EXEC in the entry block.

Noticed this in a more involved shader in a piglit run.


http://reviews.llvm.org/D18162

Files:
  lib/CodeGen/ProcessImplicitDefs.cpp
  lib/CodeGen/TwoAddressInstructionPass.cpp
  lib/Target/AMDGPU/AMDGPU.h
  lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
  lib/Target/AMDGPU/CMakeLists.txt
  lib/Target/AMDGPU/SILowerControlFlow.cpp
  lib/Target/AMDGPU/SIRegisterInfo.h
  lib/Target/AMDGPU/SIWholeQuadMode.cpp
  test/CodeGen/AMDGPU/si-scheduler.ll
  test/CodeGen/AMDGPU/wqm.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D18162.50671.patch
Type: text/x-patch
Size: 39167 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160314/eea5cab1/attachment.bin>