feat(cufile): register device buffers before use - #10
Merged
Conversation
karlowich
requested changes
Jul 7, 2026
karlowich
left a comment
karlowich
left a comment
Collaborator
There was a problem hiding this comment.
Great, we definitely need support for this.
I think I prefer the flag to be called --register_bufs, I know that flips the default behaviour, but perhaps it's fine to have the naive version by default.
Add a --register_bufs flag (opts->register_bufs) that registers the cufile backend's device read buffers with cuFileBufRegister after allocation and deregisters them before cudaFree. Off by default. Rejected for non-cufile backends, mirroring the --async validation. Signed-off-by: Frederik Madsen <f.madsen@samsung.com> Assisted-by: Claude:claude-opus-4-8
fmma
force-pushed
the
fix/cufile-buf-register
branch
from
July 8, 2026 06:20
ff4ce66 to
2cf8feb
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Register the cufile backend's device read buffers with cuFileBufRegister after allocation and deregister them before cudaFree.
Add a --no-buf-register flag (opts->no_buf_register) to skip registration. Rejected for non-cufile backends, mirroring the --async validation.
Assisted-by: Claude:claude-opus-4-8