[Fusion] Enhance the prologue epilogue fusion #277
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Previously, to enable automatic prologue epilogue fusion, we require the template written in a single kernel function (e.g., cuda kernel or cpu kernel), and the prologue/epilogue tensors must be directly used in the kernel function. This PR allows the user to define launch function and use arbitrary implementation (e.g., call multiple kernel functions, use workspace, ...) while still support fusion. This enhancement is necessary to support parallel-k optimization in the operator-level.