[all-commits] [llvm/llvm-project] f2c25c: [CostModel] remove cost-kind predicate for some ve...

RotateRight via All-commits all-commits at lists.llvm.org
Sun Oct 25 12:21:54 PDT 2020


  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: f2c25c70791de95d2466e09b5b58fc37f6ccd7a4
      https://github.com/llvm/llvm-project/commit/f2c25c70791de95d2466e09b5b58fc37f6ccd7a4
  Author: Sanjay Patel <spatel at rotateright.com>
  Date:   2020-10-25 (Sun, 25 Oct 2020)

  Changed paths:
    M llvm/include/llvm/CodeGen/BasicTTIImpl.h
    M llvm/test/Analysis/CostModel/ARM/intrinsic-cost-kinds.ll
    M llvm/test/Analysis/CostModel/ARM/reduce-add.ll
    M llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll

  Log Message:
  -----------
  [CostModel] remove cost-kind predicate for some vector reduction costs

This is a modified 2nd try of 22d10b8ab44f
(reverted by 1c8371692d because it managed
to expose an existing crashing bug that should be fixed by
74ffc823 ).

Original commit message:

This is similar in spirit to 01ea93d85d6e (memcpy) except that
here the underlying caller assumptions were created for vectorizer
use (throughput) rather than other passes.

That meant targets could have an enormous throughput cost with no
corresponding size, latency, or blended cost increase.
The ARM costs show a small difference between throughput and
size because there's an underlying difference in cmp/sel
costs that is also predicated on cost-kind.

Paraphrasing from the previous commits:
This may not make sense for some callers, but at least now the
costs will be consistently wrong instead of mysteriously wrong.

Targets should provide better overrides if the current modeling
is not accurate.




More information about the All-commits mailing list