Repository navigation
Expose best-effort native memory reclamation - #116
Merged
Merged
Conversation
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Android allocators can retain unused native pages after inference or model close. Add explicit
SpeechMemory.releaseUnused()backed by bionicmallopt(M_PURGE, 0), returning allocator support without closing models or changing allocator policy. Consumers own the idle scheduling and should call it on a worker thread.Add a device regression that purges between repeated Parakeet decodes and verifies identical text and word times, then exercises repeated reclamation after close. Document the API and version 0.0.27 in all ten README languages.
Validation: 184 SDK JVM tests passed; 13 selected connected SDK tests passed on an arm64 Samsung SM-S918B, Android 16 / API 36, including the new live-model preservation case, Nemotron streaming, Sortformer and ReDimNet. Release and debug native packages build on arm64. No model weights, model pins or inference algorithms changed.