[PATCH] D37605: AMDGPU: Match load d16 hi instructions

[Matt Arsenault via Phabricator via llvm-commits]

arsenm created this revision.
Herald added subscribers: t-tye, tpr, dstuttard, yaxunl, nhaehnle, wdng, kzhuravl.

Also starts selecting global loads for constant address
in some cases. Some end up selecting to mubuf still, which
requires investigation.

      

We still get sub-optimal regalloc and extra waitcnts inserted
due to not really tracking the liveness of the separate register
halves.


https://reviews.llvm.org/D37605

Files:
  lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
  lib/Target/AMDGPU/AMDGPUInstructions.td
  lib/Target/AMDGPU/BUFInstructions.td
  lib/Target/AMDGPU/DSInstructions.td
  lib/Target/AMDGPU/FLATInstructions.td
  test/CodeGen/AMDGPU/extract_vector_elt-i16.ll
  test/CodeGen/AMDGPU/fabs.f16.ll
  test/CodeGen/AMDGPU/load-hi16.ll
  test/CodeGen/AMDGPU/packed-op-sel.ll
  test/CodeGen/AMDGPU/sext-in-reg.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D37605.114288.patch
Type: text/x-patch
Size: 42462 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170908/ea0f1dc9/attachment.bin>