[all-commits] [llvm/llvm-project] b30d53: [AArch64] Add better fcmp costs for expanded predi...

[David Green via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: b30d5315b7ed6d9e2f5aab7140896d16ed020a40
      https://github.com/llvm/llvm-project/commit/b30d5315b7ed6d9e2f5aab7140896d16ed020a40
  Author: David Green <david.green at arm.com>
  Date:   2025-08-04 (Mon, 04 Aug 2025)

  Changed paths:
    M llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
    M llvm/test/Analysis/CostModel/AArch64/cmp.ll
    M llvm/test/Analysis/CostModel/AArch64/fcmp.ll
    M llvm/test/Analysis/CostModel/AArch64/sve-cmpsel.ll
    M llvm/test/Analysis/CostModel/AArch64/sve-fcmp.ll
    M llvm/test/Analysis/CostModel/AArch64/vector-select.ll
    M llvm/test/Transforms/SLPVectorizer/scalarazied-result.ll

  Log Message:
  -----------
  [AArch64] Add better fcmp costs for expanded predicates (#147940)

Certain fcmp predicates need to be expanded into multiple operations and
or'd together. This adds some more accurate cost modelling for them
based on the predicate. Unsupported operations are given the cost of a
libcall and the latency is set to 2 as that seemed to be fairly common
between different CPUs.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications