[Mlir-commits] [mlir] [mlir][x86vector] AVX2 i8/i32 Dot Op (PR #147908)

[Adam Siemieniuk]

================
@@ -420,6 +420,57 @@ def DotOp : AVX_LowOp<"dot", [Pure,
   }];
 }
 
+//----------------------------------------------------------------------------//
+// AVX Int8 Dot
----------------
adam-smnk wrote:

Let's lean more into its intended `i8 * i8 -> i32` semantics and align closer to the dot bf16.
I'd suggest renaming the op to `dot.i8` and make the op inputs take appropriately sized vectors of `i8`.

The inputs can be packed into `i32` as a part of `getIntrinsicOperands` through bitcasts to align with the underlying intrinsic API.

https://github.com/llvm/llvm-project/pull/147908