[all-commits] [llvm/llvm-project] 7f0850: Introduce `arith.scaling_extf` and `arith.scaling_...

[Umang Yadav via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 7f08503a3bf3acdd2a58ac712d5e95682ce583dd
      https://github.com/llvm/llvm-project/commit/7f08503a3bf3acdd2a58ac712d5e95682ce583dd
  Author: Umang Yadav <29876643+umangyadav at users.noreply.github.com>
  Date:   2025-06-09 (Mon, 09 Jun 2025)

  Changed paths:
    M mlir/include/mlir/Dialect/Arith/IR/ArithOps.td
    M mlir/include/mlir/Dialect/Arith/Transforms/Passes.h
    M mlir/include/mlir/IR/Builders.h
    M mlir/lib/Dialect/Arith/IR/ArithOps.cpp
    M mlir/lib/Dialect/Arith/Transforms/ExpandOps.cpp
    M mlir/lib/IR/Builders.cpp
    M mlir/test/Dialect/Arith/expand-ops.mlir

  Log Message:
  -----------
  Introduce `arith.scaling_extf` and `arith.scaling_truncf` (#141965)

This PR adds `arith.scaling_truncf` and `arith.scaling_extf` operations
which supports the block quantization following OCP MXFP specs listed
here
https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf

OCP MXFP Spec comes with reference implementation here
https://github.com/microsoft/microxcaling/tree/main

Interesting piece of reference code is this method `_quantize_mx`
https://github.com/microsoft/microxcaling/blob/7bc41952de394f5cc5e782baf132e7c7542eb4e4/mx/mx_ops.py#L173.

Both `arith.scaling_truncf` and `arith.scaling_extf` are designed to be
an elementwise operation. Please see description about them in
`ArithOps.td` file for more details.
 
Internally, 

`arith.scaling_truncf` does the
`arith.truncf(arith.divf(input/(2^scale)))`. `scale` should have
necessary broadcast, clamping, normalization and NaN propagation done
before callling into `arith.scaling_truncf`.

`arith.scaling_extf` does the `arith.mulf(2^scale, input)` after taking
care of necessary data type conversions.


CC: @krzysz00 @dhernandez0 @bjacob @pashu123 @MaheshRavishankar
@tgymnich

---------

Co-authored-by: Prashant Kumar <pk5561 at gmail.com>
Co-authored-by: Krzysztof Drewniak <Krzysztof.Drewniak at amd.com>



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications