Use direct evaluation of kernel functions on GPU #39
Conversation
This file contains 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
Atomic add on shared memory is very slow. Is it Atomix's fault?
Much faster!
Doesn't change performance.
It is the same or faster than polynomial approximation (and more accurate!). Especially for shared-memory interpolation, it seems to improve performance by a lot.
Doesn't really affect performance, on GPU at least.
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #39 +/- ##
=======================================
Coverage 92.31% 92.32%
=======================================
Files 18 19 +1
Lines 1614 1654 +40
=======================================
+ Hits 1490 1527 +37
- Misses 124 127 +3 ☔ View full report in Codecov by Sentry. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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 seems to be slightly faster (and uses far fewer GPU registers) than approximate evaluations based on piecewise polynomials.
Also, specifically on CUDA, we now default to
kernel = KaiserBesselKernel()instead ofBackwardsKaiserBesselKernel()as it seems to be a bit faster. Accuracy is not affected since both kernels have equivalent precisions.