[PATCH] D54042: [AMDGPU] Extend the SI Load/Store optimizer to combine more things.

[Neil Henning via Phabricator via llvm-commits]

sheredom created this revision.
sheredom added reviewers: nhaehnle, arsenm.
sheredom added a project: AMDGPU.
Herald added subscribers: llvm-commits, t-tye, tpr, dstuttard, yaxunl, wdng, jvesely, kzhuravl.

I've extended the load/store optimizer to be able to produce dwordx3 loads and stores, and also enable it to produce dwordx8 and dwordx16 sgpr loads. This change allows many more load/stores to be combined, and results in much more optimal code for our hardware.


Repository:
  rL LLVM

https://reviews.llvm.org/D54042

Files:
  lib/Target/AMDGPU/SILoadStoreOptimizer.cpp
  test/CodeGen/AMDGPU/cvt_f32_ubyte.ll
  test/CodeGen/AMDGPU/early-if-convert-cost.ll
  test/CodeGen/AMDGPU/insert_vector_elt.ll
  test/CodeGen/AMDGPU/llvm.amdgcn.buffer.load.ll
  test/CodeGen/AMDGPU/llvm.amdgcn.buffer.store.ll
  test/CodeGen/AMDGPU/llvm.amdgcn.s.buffer.load.ll
  test/CodeGen/AMDGPU/merge-stores.ll
  test/CodeGen/AMDGPU/store-global.ll
  test/CodeGen/AMDGPU/store-v3i64.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D54042.172406.patch
Type: text/x-patch
Size: 62993 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20181102/1e67b5ba/attachment.bin>