Vectorize pooling for optimization #2905
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Outdated suggestions cannot be applied.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.
Pull Request Template
Checklist
run-checks all
script has been executed.Changes
Uses the custom transpose kernel from implicit GEMM convolution to switch all pooling operations to NHWC, and vectorizes along the channel dimension. This improves performance by approx. 20-25% when processing contiguous NCHW tensors, more when processing contiguous NHWC (i.e. output from implicit GEMM convolution). While migrating the kernels, I also renamed all variables that were just numbered to something more semantically meaningful.
Testing
All tests pass, and changes to the kernels are kept as minimal as possible.