[all-commits] [llvm/llvm-project] fdf1fd: [libc][math] Improved ExhaustiveTest performance.

[orex via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: fdf1fda5a8c8f12d51981da1f0f457a6664c4bc5
      https://github.com/llvm/llvm-project/commit/fdf1fda5a8c8f12d51981da1f0f457a6664c4bc5
  Author: Kirill Okhotnikov <okir at google.com>
  Date:   2022-07-01 (Fri, 01 Jul 2022)

  Changed paths:
    M libc/test/src/math/exhaustive/exhaustive_test.cpp
    M libc/test/src/math/exhaustive/exhaustive_test.h
    M libc/test/src/math/exhaustive/exp2f_test.cpp
    M libc/test/src/math/exhaustive/expf_test.cpp
    M libc/test/src/math/exhaustive/expm1f_test.cpp
    M libc/test/src/math/exhaustive/hypotf_test.cpp
    M libc/test/src/math/exhaustive/log10f_test.cpp
    M libc/test/src/math/exhaustive/log1pf_test.cpp
    M libc/test/src/math/exhaustive/log2f_test.cpp
    M libc/test/src/math/exhaustive/logf_test.cpp

  Log Message:
  -----------
  [libc][math] Improved ExhaustiveTest performance.

Previous implementation splits value ranges around threads. Because of
very different performance of testing functions over different ranges,
CPU utilization were poor. Current implementation split test range
over small pieces and threads take the pieces when they finish with
previous. Therefore the CPU load is constant during testing.

Differential Revision: https://reviews.llvm.org/D128995