[all-commits] [llvm/llvm-project] 71dc3d: [ARM] Improve min/max vector reductions on Arm

[Caleb Zulawski via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 71dc3de533b9247223c083a3b058859c9759099c
      https://github.com/llvm/llvm-project/commit/71dc3de533b9247223c083a3b058859c9759099c
  Author: Caleb Zulawski <caleb.zulawski at gmail.com>
  Date:   2023-03-22 (Wed, 22 Mar 2023)

  Changed paths:
    M llvm/lib/Target/ARM/ARMISelLowering.cpp
    A llvm/test/CodeGen/ARM/vecreduce-minmax.ll

  Log Message:
  -----------
  [ARM] Improve min/max vector reductions on Arm

This patch adds some more efficient lowering for vecreduce.min/max under NEON,
using sequences of pairwise vpmin/vpmax to reduce to a single value.

This nearly resolves issues such as #50466, #40981, #38190.

Differential Revision: https://reviews.llvm.org/D146404