Vectorization of convolution functions in utils/convolve.py #72
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
|
@FMarmoreo could you test whether the output of the updated functions are all equivalent to the previous versions and post the results? Otherwise, a change in behavior would need to be documented. |
|
Here a test about the new implementation of phase_conv2d, compared to the old implementation (after having fixed the bug as in PR #66): |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.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.
The implementation of the phase_conv2d, nan_conv2d and norm_angle has been replaced by a vectorized implementation approach to significantly speed up code execution (~200x faster for nan_conv2d and ~40x faster for phase_conv2d).
WARNING: This implementation also fixes the boundary conditions issue taken into account by PR #66.
Please consider to merge PR #66 before this one for result consistency.