[Mlir-commits] [mlir] [mlir][AMDGPU] Add scaled floating point conversion ops fp8 (PR #141554)

[Krzysztof Drewniak]

================
@@ -112,6 +112,38 @@ def AMDGPU_ExtPackedFp8Op :
   }];
 }
 
+def AMDGPU_ScaledExtPackedOp
+    : AMDGPU_Op<"scaled_ext_packed", [Pure]>,
+      Arguments<(
+          ins AnyTypeOf<[VectorOfLengthAndType<[2, 3, 4], [F8E5M2, F8E4M3FN]>,
----------------
krzysz00 wrote:

That conversion isn't packed in the input - it only converts on single bf8/fp8, but it does still have to update one half of a v2f16 with the result

https://github.com/llvm/llvm-project/pull/141554