Implement abstract per-GPU cache helper class - #814
Merged
Conversation
meta-cla
Bot
added
the
CLA Signed
NicolasHug
reviewed
Aug 14, 2025
scotts
reviewed
Aug 15, 2025
scotts
reviewed
Aug 15, 2025
scotts
reviewed
Aug 15, 2025
Contributor
|
@dvrogozh, thank you for working on this! These abstractions are very much needed. The small tweaks to the implementation we talked about should be enough. |
This commit introduces abstract per-GPU cache helper class. It allows to store and reuse HW contexts which are expensive to create. Class is abstract and allows storing different object types. Co-authored-by: Nicolas Hug <contact@nicolas-hug.com> Signed-off-by: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
Signed-off-by: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
Signed-off-by: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
Collaborator
Author
These ci failures seem unrelated. |
Contributor
|
@dvrogozh, yup, agreed, it's a problem we know about and should start tracking. I'm going to merge. |
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This commit introduces per-GPU cache helper class based on original design of the cache in CUDA device interface. New helper allows to store and reuse HW contexts which are expensive to create. Class is templated and allows storing different object types. At the moment it's applied to cache CUDA ffmpeg device contexts (
AVBufferRef) and NPP contexts (NppStreamContext). Class can be further reused in non-CUDA device implementations.CC: @scotts, @NicolasHug, @eromomon