[all-commits] [llvm/llvm-project] 9294a1: [mlir][tosa] Rework tosa.apply_scale lowering for ...

[Rob Suderman via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 9294a1e9a8ab8fed60ee1b0737e944077b5339cd
      https://github.com/llvm/llvm-project/commit/9294a1e9a8ab8fed60ee1b0737e944077b5339cd
  Author: Robert Suderman <suderman at google.com>
  Date:   2022-05-17 (Tue, 17 May 2022)

  Changed paths:
    M mlir/include/mlir/Conversion/Passes.td
    M mlir/include/mlir/Conversion/TosaToArith/TosaToArith.h
    M mlir/lib/Conversion/TosaToArith/TosaToArith.cpp
    M mlir/lib/Conversion/TosaToArith/TosaToArithPass.cpp
    M mlir/test/Conversion/TosaToArith/tosa-to-arith.mlir

  Log Message:
  -----------
  [mlir][tosa] Rework tosa.apply_scale lowering for 32-bit

Added handling rounding behavior in 32-bits for when possible. This
avoids kernel compilation generating scalarized code on platforms where
64-bit vectors are not available.

As the 48-bit lowering requires 64-bit anyway, we added a full 64-bit
solution simplifying the old path.

Reviewed By: dcaballe, mravishankar

Differential Revision: https://reviews.llvm.org/D125583