[llvm-branch-commits] [mlir] [mlir][linalg] Add transform operator for Winograd Conv2D algorithm (PR #96182)
Oleksandr Alex Zinenko via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Fri Jun 21 08:24:26 PDT 2024
================
@@ -2587,4 +2587,55 @@ def MapCopyToThreadsOp :
}];
}
+//===----------------------------------------------------------------------===//
+// Winograd Conv2D
+//===----------------------------------------------------------------------===//
+
+def WinogradConv2DOp : Op<Transform_Dialect,
+ "structured.winograd_conv2d",
+ [FunctionalStyleTransformOpTrait, MemoryEffectsOpInterface,
+ TransformOpInterface, TransformEachOpTrait,
+ ReportTrackingListenerFailuresOpTrait]> {
+ let description = [{
+ Winograd Conv2D algorithm will convert linalg Conv2D operator into batched
+ matrix multiply. Before the matrix multiply, it will convert filter and
+ input into a format suitable for batched matrix multiply. After the matrix
+ multiply, it will convert output to the final result tensor.
+
+ The algorithm F(m x m, r x r) is
+
+ Y = A^T x [(G x g x G^T) @ (B^T x d x B)] x A
+
+ The size of output Y is m x m. The size of filter g is r x r. The size of
+ input d is (m + r - 1) x (m + r - 1). A^T, A, G^T, G, B^T, and B are
+ transformation matrices.
+
+ #### Return modes:
+
+ This operation fails if `target` is unsupported. Otherwise, the operation
----------------
ftynse wrote:
```suggestion
This operation produces a silenceable failure if `target` is unsupported. Otherwise, the operation
```
https://github.com/llvm/llvm-project/pull/96182
More information about the llvm-branch-commits
mailing list