[all-commits] [llvm/llvm-project] 0b8028: [NVPTX] Preserve v16i8 vector loads when legalizing

[Pierre-Andre Saulais via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 0b80288e9e0b12f9680d9f2cfdff5686c38982d2
      https://github.com/llvm/llvm-project/commit/0b80288e9e0b12f9680d9f2cfdff5686c38982d2
  Author: Pierre-Andre Saulais <pierre-andre at codeplay.com>
  Date:   2023-10-19 (Thu, 19 Oct 2023)

  Changed paths:
    M llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
    M llvm/test/CodeGen/NVPTX/LoadStoreVectorizer.ll

  Log Message:
  -----------
  [NVPTX] Preserve v16i8 vector loads when legalizing

This is done by lowering v16i8 loads into LoadV4 operations with i32
results instead of letting ReplaceLoadVector split it into smaller
loads during legalization. This is done at dag-combine1 time, so that
vector operations with i8 elements can be optimised away instead of
being needlessly split during legalization, which involves storing to
the stack and loading it back.