Skip to content

[GPU] Clean oneDNN cache #33071

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
p-durandin wants to merge 6 commits into
base: master
Choose a base branch
from
Open

[GPU] Clean oneDNN cache #33071

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
0ba48ad
[GPU] Clean oneDNN cache
p-durandin Nov 28, 2025
e29dd00
[GPU] Clean oneDNN cache
p-durandin Dec 5, 2025
405785c
[GPU] Clean oneDNN cache
p-durandin Dec 5, 2025
eee7dc5
[GPU] Clean oneDNN cache
p-durandin Dec 5, 2025
674cae9
[GPU] Clean oneDNN cache
p-durandin Dec 5, 2025
5728032
[GPU] Clean oneDNN cache
p-durandin Dec 5, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 21 additions & 0 deletions samples/cpp/benchmark_app/main.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -329,6 +329,25 @@ void fuse_mean_scale(ov::preprocess::PrePostProcessor& preproc, const benchmark_
}
} // namespace

/**
* Sentry class to ensure the resource releasing from CompiledModel.
* Release static global oneDNN cache used for inference with release_memory
* method. To be called once at the end of inference.
*/
class ModelSentry {
public:
ModelSentry(ov::CompiledModel& compiledModel) : _model(compiledModel) {}

~ModelSentry() {
if (_model) {
_model.release_memory();
}
}

private:
ov::CompiledModel _model;
};

/**
* @brief The entry point of the benchmark application
*/
Expand All @@ -337,6 +356,8 @@ int main(int argc, char* argv[]) {
try {
ov::CompiledModel compiledModel;

ModelSentry releaser(compiledModel);
Copy link
Contributor

@Wovchena Wovchena Dec 5, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does it actually capture the model? Since _model is a copy now, there's a risk that ModelSentry does nothing. main() should use _model instead of compiledModel to fix that


// ----------------- 1. Parsing and validating input arguments
// -------------------------------------------------
next_step();
Expand Down
Loading