OPT: eliminating temp buffer in cascading joins #3145
Merged
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
9prady9
reviewed
Jun 21, 2021
There was a problem hiding this comment.
Change looks simple and good. However, many linux CUDA/OpenCL tests failed. Did all the tests pass for your local build ? @willyborn
It is faster to join multiple array's directly into the final buffer, iso using temp buffers. Previous flow: - join (array A & array B) into temp buffer - join (temp & array C) into final buffer New flow: - join (array A, array B & array C) into final buffer
b8b5fbb to
2398993
Compare
|
All tests passed.
At the introduction, I ran all tests OpenCL, CUDA & CPU, (windows 10
fftw3&openblas with MSVC) with satisfactory results.
I checked the logs of the failures, and concluded that they all were a
result of bad downloads (timeouts) during installation.
To be certain, I performed all the tests again. The master has changed
too much in the meantime, so that *I had to rebase again*.
The same conclusion as before, although test_svd_dense_opencl failed with
"LAPACKE Error (-5)".
Exactly as christopher gee reported.
I believe this has nothing to do with the changes with this PR.
I will put my findings in the thread of christophe_gee. (#3147)
Willy
…On Mon, 21 Jun 2021 at 10:04, pradeep ***@***.***> wrote:
***@***.**** commented on this pull request.
Change looks simple and good. However, many linux CUDA/OpenCL tests
failed. Did all the tests pass for your local build ? @willyborn
<https://github.com/willyborn>
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3145 (review)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AQ2WGPDNHLDHRALE4H5EWALTT3XA5ANCNFSM46MUBNWQ>
.
|
9prady9
approved these changes
Jun 22, 2021
There was a problem hiding this comment.
@willyborn That is fine. I just wanted to make sure failures aren't related to the change by any chance.
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.
It is faster to join multiple array's directly into the final buffer, iso using temp buffers with cascading joins
Previous flow:
New flow:
Performance improvement for gray2rgb & ycbr2rgb are:
When combined with PR#3144 (join optimization)
Description
Performance improvement for gray2rgb & ycbr2rgb functions.
Less memory consumption, since 1 intermediate buffer is eliminated by this PR, and 3 extra intermediate buffers by PR3144.
Changes to Users
No functional changes.
Checklist