[all-commits] [llvm/llvm-project] fbe739: [mlir][x86] Fix - replace `vector.load` with `vect...

[Arun Thangamani via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: fbe7392bdf453cff70615c242d74f11789dd76f1
      https://github.com/llvm/llvm-project/commit/fbe7392bdf453cff70615c242d74f11789dd76f1
  Author: Arun Thangamani <arun.thangamani at intel.com>
  Date:   2026-06-04 (Thu, 04 Jun 2026)

  Changed paths:
    M mlir/lib/Dialect/X86/Transforms/VectorContractToAMXDotProduct.cpp
    M mlir/lib/Dialect/X86/Transforms/VectorContractToPackedTypeDotProduct.cpp
    M mlir/test/Dialect/X86/AMX/vector-contract-to-tiled-dp.mlir
    M mlir/test/Dialect/X86/vector-contract-to-packed-type-dotproduct.mlir

  Log Message:
  -----------
  [mlir][x86] Fix - replace `vector.load` with `vector.transfer_read` (#201503)

This patch replaces `vector.load` with `vector.transfer_read` to load
elements as `vector<2x32xi8>` instead of `vector<64xi8>`. This fixes the
online shuffling for `int8` and `f8` types.



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