[Mlir-commits] [mlir] [mlir][x86vector] Lower vector.contract to sequence of FMAs (PR #163382)

[Md Asghar Ahmad Shahid]

shahidact wrote:

> A high-level concern of mine is that this transform requires really specific IR to match at all. It might be difficult to use it in a larger schedule.

+1

https://github.com/llvm/llvm-project/pull/163382