Replace direct cudaMemcpyAsync calls with utility functions (within /include)
#17557
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
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
vuule
added
improvement
davidwendt
reviewed
Dec 10, 2024
…remove-memcpy-include
…into remove-memcpy-include
vuule
changed the title
cudaMemcpyAsync calls with utility functions (within /include)
davidwendt
reviewed
Dec 10, 2024
Co-authored-by: David Wendt <[email protected]>
davidwendt
approved these changes
Dec 10, 2024
nvdbaranec
approved these changes
Dec 10, 2024
Comment on lines
52
to
53
| return cudf::detail::make_host_vector_sync( | ||
| device_span<T const>{col_view.data<T>() + element_index, 1}, stream).front(); |
There was a problem hiding this comment.
Quibble: In some alternate universe where pinned memory isn't being used and we drop back to the pageable resource, this is probably somewhat worse than what's there now. I imagine that's not a case we really care about though.
There was a problem hiding this comment.
That's true. FWIW, we did a full benchmark run with a change like this in the device_scalar, and there was no negative impact from involving the heap unnecessarily.
karthikeyann
approved these changes
Dec 11, 2024
|
/merge |
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Outdated suggestions cannot be applied.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.
Description
Replaced the calls to
cudaMemcpyAsyncwith the newcuda_memcpy/cuda_memcpy_asyncutility, which optionally avoids using the copy engine.Also took the opportunity to use
cudf::detail::host_vectorand its factories to enable wider pinned memory use.Checklist