[code-reflection] RFR: Model lifetimes of onnx session-related objects more explicitly
Maurizio Cimadamore
mcimadamore at openjdk.org
Mon Mar 3 11:27:07 UTC 2025
On Fri, 28 Feb 2025 12:42:24 GMT, Maurizio Cimadamore <mcimadamore at openjdk.org> wrote:
> The class representing an onnx session is auto closeable. But, in the current code, a session is closed immediately after its `run` method is called. This is problematic because a session returns some ORTValues (tensors) which also need to be freed, but that cannot be freed immediately after calling `run` (as they need to be used by clients).
>
> To address this problem, I tweaked the session code to accept an external arena. All the allocation of session-related data structures now happens using that external arena. This means that the client can now be in charge of managing the lifetime of a session (see changes to MNIST demo).
>
> To test, I tweaked the MNIST code to do 10K iterations on each button pressed. Predictably, a single button pressed resulted in over 3g of memory being leaked. With these changes the memory arrives at ~400K (there is still some minor leak, but not sure worth pushing more).
>
> If the changes to the demo are not deemed good, I can withdraw this PR -- I mostly wanted to capture the result of my exploration somewhere.
Meta-comment -- it would perhaps be better to split this PR from the discussion of what a better high-level API should be? This PR really adds ways to correctly manage memory that is local to a session, so that there's no memory leaks. If we don't like having developers manually create an arena we can fix that later in a more API-oriented PR (as this PR should introduce enough machinery to implement whatever API is desirable on top).
-------------
PR Comment: https://git.openjdk.org/babylon/pull/332#issuecomment-2694054943
More information about the babylon-dev
mailing list