[CPU] Throw when release_memory is called during inference #27520
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
|
@EgorDuplensky , could you please review? |
EgorDuplensky
approved these changes
Dec 9, 2024
dmitry-gorokhov
approved these changes
Dec 16, 2024
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Outdated suggestions cannot be applied.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.
Details:
This PR changes the behavior of the
CompiledModel::release_memory()implementation in the CPU plugin for the situation when the method is being called concurrently with the other graph state modifying methods (e.g. graph initialization, inference, properties request). This is necessary to ensure thread safety and provide a clear defined behavior when the method is called concurrently.Also, the PR contains some refactoring of the Infer request implementation, aimed at decoupling the InferRequest implementation from the compiled model internals and providing a safer interface that ensures thread safe access to the CPU graph structures.