[all-commits] [llvm/llvm-project] 517061: [AArch64] turn extended vecreduce bigger than v16i...

[Zain Jaffal via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 5170610b5789cde77a948fe57a715c512dcfe350
      https://github.com/llvm/llvm-project/commit/5170610b5789cde77a948fe57a715c512dcfe350
  Author: Zain Jaffal <z_jaffal at apple.com>
  Date:   2023-01-31 (Tue, 31 Jan 2023)

  Changed paths:
    M llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
    M llvm/test/CodeGen/AArch64/neon-dotreduce.ll

  Log Message:
  -----------
  [AArch64] turn extended vecreduce bigger than v16i8 into udot/sdot

We can do this by breaking vecreduce into v16i8 vectors generating udot/sdot and concatenating them.

Differential Revision: https://reviews.llvm.org/D141693