[Mlir-commits] [mlir] [mlir][vector] Deal with special patterns when emulating masked load/store (PR #75587)

[Hsiangkai Wang]

Hsiangkai wrote:

> +1 to what @kuhar suggested. I think the only missing pattern is to fold:
> 
> ```
>   %0 = vector.constant_mask [2] : vector<4xi1>
>   %1 = vector.extract %0[0] : i1 from vector<4xi1>
> ```
> 
> into the specific constant value:
> 
> ```
>   %0 = arith.constant 1 : i1
> ```
> 
> The rest of the folding that you are looking for should happen automatically after that.

Thanks for your comments. I will work on the missing fold pattern.

https://github.com/llvm/llvm-project/pull/75587