[PATCH] D79898: [AMDGPU] Optimized indirect multi-VGPR addressing

[Stanislav Mekhanoshin via Phabricator via llvm-commits]

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

SelectMOVRELOffset prevents peeling of a constant from an index
if final base could be negative. isBaseWithConstantOffset() succeeds
if a value is an "add" or "or" operator. In case of "or" it shall
be an add-like "or" which never changes a sign of the sum. I.e. we
can safely allow peeling if operator is an "or".


https://reviews.llvm.org/D79898

Files:
  llvm/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
  llvm/test/CodeGen/AMDGPU/extract_vector_dynelt.ll
  llvm/test/CodeGen/AMDGPU/indirect-addressing-si.ll
  llvm/test/CodeGen/AMDGPU/insert_vector_dynelt.ll
  llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D79898.263809.patch
Type: text/x-patch
Size: 7886 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20200513/3870da7c/attachment-0001.bin>