Pad gpu kernel support and test#40
Open
harz05 wants to merge 2 commits into
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR implements the gpu support for the Pad operator (constant mode, matching the CPU op which also supports only constant mode).
Approach: one thread per output element (gather). Each thread decomposes its flat output index into per-dim coordinates, tests whether they map back inside the input (interior vs padding) and either copies the input element or writes the constant. Single pass, no pre-fill. Shapes/strides/pads are baked in as literals at codegen time. One thread per input element could have been used too but from what I can understand it will require 2 kernels, one that fills output with constant and one to scatter each input element into its output position
Test verified on Colab T4, existing Pad.onnx has been use and expected output generated from numpy.pad