[Mlir-commits] [mlir] [mlir][linalg] Implement Winograd Conv2D. (PR #94470)

[Hsiangkai Wang]

================
@@ -154,4 +154,96 @@ def Linalg_SoftmaxOp : Linalg_Op<"softmax",
   let hasVerifier = 1;
 }
 
+def Linalg_WinogradFilterTransformOp : Linalg_Op<"winograd_filter_transform",
+    [DeclareOpInterfaceMethods<TilingInterface,
+     ["getIterationDomain",
+      "getLoopIteratorTypes",
+      "getResultTilePosition",
+      "getTiledImplementation"]>]> {
+  let summary = "Winograd filter transform operator";
+  let description = [{
+    linalg.winograd_filter_transform transforms the filter of conv2D.
+  }];
+
+  let arguments = (ins AnyRankedTensor:$filter,
+                       AnyRankedTensor:$output,
+                       I64Attr:$output_height,
+                       I64Attr:$output_width,
+                       I64Attr:$m,
+                       I64Attr:$r
----------------
Hsiangkai wrote:

`m` and `r` are variable names in the Winograd algorithm. I added more description for the operators.

https://github.com/llvm/llvm-project/pull/94470