[all-commits] [llvm/llvm-project] 929eb5: [mlir] Rewrites for I2 to I8 signed and unsigned e...

[ziereis via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 929eb500d4c9b3fff0693c49fd55c8093dc1ad62
      https://github.com/llvm/llvm-project/commit/929eb500d4c9b3fff0693c49fd55c8093dc1ad62
  Author: ziereis <44057120+ziereis at users.noreply.github.com>
  Date:   2025-01-15 (Wed, 15 Jan 2025)

  Changed paths:
    M mlir/lib/Dialect/Vector/Transforms/VectorEmulateNarrowType.cpp
    M mlir/test/Dialect/Vector/vector-rewrite-narrow-types.mlir

  Log Message:
  -----------
  [mlir] Rewrites for I2 to I8 signed and unsigned extension (#121298)

Adds rewrites for i2 to i8 signed and unsigned extension, similar to the
ones that already exist for i4 to i8 conversion.

I use this for i6 quantized models, and this gives me roughly a 2x
speedup for an i6 4096x4096 dequantization-matmul on an AMD 5950x.

I didn't add the rewrite for i8 to i2 truncation because I currently
don't use it, but if this is needed, I can add it as well.

---------

Co-authored-by: Andrzej Warzyński <andrzej.warzynski at gmail.com>



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