[all-commits] [llvm/llvm-project] 13d136: A better profi rebalancer

[spupyrev via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 13d1364a344328145b4bc9449fa6aae34f300886
      https://github.com/llvm/llvm-project/commit/13d1364a344328145b4bc9449fa6aae34f300886
  Author: spupyrev <spupyrev at fb.com>
  Date:   2022-01-18 (Tue, 18 Jan 2022)

  Changed paths:
    M llvm/lib/Transforms/Utils/SampleProfileInference.cpp
    A llvm/test/Transforms/SampleProfile/Inputs/profile-inference-rebalance-large.prof
    A llvm/test/Transforms/SampleProfile/profile-inference-rebalance-large.ll
    M llvm/test/Transforms/SampleProfile/profile-inference-rebalance.ll

  Log Message:
  -----------
  A better profi rebalancer

This is an extension of **profi** post-processing step that rebalances counts
in CFGs that have basic blocks w/o probes (aka "unknown" blocks). Specifically,
the new version finds many more "unknown" subgraphs and marks more "unknown"
basic blocks as hot (which prevents unwanted optimization passes).

I see up to 0.5% perf on some (large) binaries, e.g., clang-10 and gcc-8.

The algorithm is still linear and yields no build time overhead.