[all-commits] [llvm/llvm-project] fa48ff: [CodeGen] Fix neutral value of vecreduce fadd in t...

[Nikita Popov via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: fa48ff3fc981783483bcf609798fbb48524300ba
      https://github.com/llvm/llvm-project/commit/fa48ff3fc981783483bcf609798fbb48524300ba
  Author: Nikita Popov <nikita.ppv at gmail.com>
  Date:   2020-10-29 (Thu, 29 Oct 2020)

  Changed paths:
    M llvm/docs/LangRef.rst
    M llvm/test/CodeGen/AArch64/vecreduce-fadd-legalization-strict.ll
    M llvm/test/CodeGen/AArch64/vecreduce-fadd-legalization.ll
    M llvm/test/CodeGen/AArch64/vecreduce-fadd.ll
    M llvm/test/CodeGen/ARM/vecreduce-fadd-legalization-soft-float.ll
    M llvm/test/CodeGen/ARM/vecreduce-fadd-legalization-strict.ll
    M llvm/test/CodeGen/X86/vector-reduce-fadd.ll

  Log Message:
  -----------
  [CodeGen] Fix neutral value of vecreduce fadd in tests (NFC)

The neutral value is -0.0, not 0.0. This doesn't matter for "fast"
reductions due to nsz, but does matter for reassoc-only and seq
reductions.

Change tests to mostly use -0.0 where the neutral value was intended,
and add some additional test coverage in some places. Also update
LangRef to use the right value.