[all-commits] [llvm/llvm-project] 8a9d48: [mlir] Clean-up math -> libm/llvm conversion.

[Alexander Belyaev via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 8a9d4895df780231a14a1afc44e18b1f6b7eab93
      https://github.com/llvm/llvm-project/commit/8a9d4895df780231a14a1afc44e18b1f6b7eab93
  Author: Alexander Belyaev <pifon at google.com>
  Date:   2023-02-21 (Tue, 21 Feb 2023)

  Changed paths:
    M mlir/include/mlir/Conversion/MathToLLVM/MathToLLVM.h
    M mlir/include/mlir/Conversion/MathToLibm/MathToLibm.h
    M mlir/include/mlir/Conversion/Passes.td
    M mlir/lib/Conversion/MathToLLVM/MathToLLVM.cpp
    M mlir/lib/Conversion/MathToLibm/MathToLibm.cpp

  Log Message:
  -----------
  [mlir] Clean-up math -> libm/llvm conversion.

At the moment, there is an optional log1pBenefit
populateMathToLibmConversionPatterns which is used to increase the priority of
the log1p->libm pattern compared to log1p->llvm pattern that approximates
log1p with precision issues. Instead, we can have a flag for the MathToLLVM
pass to enable or disable the imprecise approximation.

Differential Revision: https://reviews.llvm.org/D144450