Removing alloca on GPU #618
Open
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
|
Hello @DiamonDinoia ! Interesting stuff :) PS: @Lenoush's contract ended so she does not work on nuffts anymore |
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Outdated suggestions cannot be applied.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.
@paquiteau and @Lenoush have noticed that alloca made things slower in their benchmarks while greatly reducing memory consumption.
Details are in #570 and mind-inria/mri-nufft-benchmark#5
Instead of using opts.gpu_* to switch with the old implementation it is better to use kernel dispatch and have pre-compiled kernels for the various scenarios. As per CPU code. One less parameter that the user has to worry about and it can obtain both higher performance and low memory consumption at the same time.
@paquiteau, @Lenoush can you benchmark this branch and let us know how it fares? I could not measure a meaningful difference with my custom code.