[Mlir-commits] [mlir] [mlir][x86] Add extra patterns for AMX nano-kernel lowering (acc with zero constant) (PR #222513)

[Adam Siemieniuk]

================
@@ -547,6 +547,7 @@ module attributes {transform.with_named_sequence} {
   transform.named_sequence @__transform_main(%arg1: !transform.any_op {transform.readonly}) {
     %func = transform.structured.match ops{["func.func"]} in %arg1 : (!transform.any_op) -> !transform.any_op
     transform.apply_patterns to %func {
+      transform.apply_patterns.x86.move_accumulator_for_contract_loop
----------------
adam-smnk wrote:

`vector_contract_to_amx_dot_product` is pretty complex on its own. It's better to avoid pulling in extra patterns unless strictly necessary. Otherwise, debugging changes gets really complex.

Would it be enough to manually tweak relevant test cases?

https://github.com/llvm/llvm-project/pull/222513