Add mlx_array_detach to break graph links before free#119
Open
odilitime wants to merge 1 commit into
Open
Conversation
Expose mlx::core::array::detach() in the C API so callers can release weight buffers without leaving sibling references in shared ArrayDesc nodes. No-op when arr.ctx is null, matching mlx_array_free behavior.
Author
Test plan results (macOS arm64)Verified locally on
Note: |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Adds
mlx_array_detach()to the C API, wrapping the existingmlx::core::array::detach()method.When releasing model weights or other arrays that may still be linked in the computation graph (e.g. sibling outputs from a traced/compiled forward pass), calling
mlx_array_free()alone can leave dangling references in sharedArrayDescnodes. Detaching first clears those sibling links so the buffer can be freed safely.mlx_array_detachinmlx/c/array.hwith documentationmlx/c/array.cpparr.ctxis null, consistent withmlx_array_free_Motivation
Downstream bindings (Go via dynamic
libmlxc, Swift, etc.) that hold weight arrays across inference sessions currently need to patcharray.cpplocally or call into C++ internals to detach before free. Exposing this in the public C API removes that friction.Test plan
mlx_array_detachis exported fromlibmlxc