[all-commits] [llvm/llvm-project] 352c46: [SLP]Improve vectorization of split loads.

[Andrew V. Tischenko via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 352c46e70716061e99cae2009daddbfc78380fda
      https://github.com/llvm/llvm-project/commit/352c46e70716061e99cae2009daddbfc78380fda
  Author: Alexey Bataev <a.bataev at outlook.com>
  Date:   2021-11-12 (Fri, 12 Nov 2021)

  Changed paths:
    M llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
    M llvm/test/Transforms/SLPVectorizer/AArch64/getelementptr.ll
    M llvm/test/Transforms/SLPVectorizer/X86/schedule_budget.ll
    M llvm/test/Transforms/SLPVectorizer/X86/split-load8_2-unord.ll
    M llvm/test/Transforms/SLPVectorizer/X86/tiny-tree.ll

  Log Message:
  -----------
  [SLP]Improve vectorization of split loads.

Need to fix ther cost estimation for split loads, since we look at the
subregs already, no need to permute them, need just to estimate
subregister insert, if it is smaller than the real register. Also, using
split loads, it might be profitable already to vectorize smaller trees
with gathering of the loads.

Differential Revision: https://reviews.llvm.org/D107188