# oMLX remaining ad-hoc mx.clear_cache sites needing stream analysis > On main, `engine/base.py` `_finish_activity` still runs `(mx.synchronize(), mx.clear_cache())` on `get_mlx_executor()` after every non-streaming request, with a comment that it always clears because gating on `_active_count == 0` grew the pool without bound (issue 684). Parent: [Mac local LLMs: oMLX, Rapid-MLX and related internals](https://llms-explorer.com/tree/mac-local-llms-omlx-and-rapid-mlx-internals/) · 2 facets · 12 facts · page: https://llms-explorer.com/tree/omlx-remaining-ad-hoc-mx-clear-cache-sites-needi/ ## Facts - On main, `engine/base.py` `_finish_activity` still runs `(mx.synchronize(), mx.clear_cache())` on `get_mlx_executor()` after every non-streaming request, with a comment that it always clears because gating on `_active_count == 0` grew the pool without bound (issue 684). — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/engine/base.py) - The same bare pair remains in `engine/reranker.py`, `engine/tts.py`, `engine/stt.py` and `engine/sts.py` (one site each, via `get_mlx_executor()`), `engine/dflash.py` (two sites, in model eviction and its settle loop) and `admin/oq_manager.py` (two sites inside retry loops of 3). — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/engine/dflash.py) - `engine_pool.py` has about 18 sites using the same `(mx.synchronize(), mx.clear_cache())` lambda on the MLX executor for load, unload, emergency and settle paths; none passes a stream or takes `_mx_buffer_access_lock`. — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/engine_pool.py) - `engine/vlm.py` still calls bare `mx.synchronize(); mx.clear_cache()` after vision encoding and in the `finally` of the diffusion stream, on the calling thread. — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/engine/vlm.py) - `engine/embedding.py` calls `mx.clear_cache()` after `mx.synchronize()` only when `fairness.should_clear_cache()` is true, inside the batch worker's `finally`. — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/engine/embedding.py) - `oq.py` has about 37 `mx.clear_cache()` mentions: most follow `mx.eval` and `del` in per-tensor or per-layer quantization loops, many with a preceding `mx.synchronize()`, and one set replaces `mx.clear_cache` with a no-op in a test or fake-array harness. — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/oq.py) - In `patches/glm_moe_dsa/generate_patch.py` the decode-side clear every 512 steps is converted to `_sync_and_clear_cache(self._stream)`, but the patched prefill loop still calls bare `mx.clear_cache()` after `mx.eval` twice (lines near 164 and 172). — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/patches/glm_moe_dsa/generate_patch.py) - `speculative/vlm_mtp.py` now uses `_sync_and_clear_cache(_vlm_generation_stream)` at three sites, matching PR 2413. — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/speculative/vlm_mtp.py) - `scheduler.py` routes every clear through `Scheduler._clear_cache`, which calls `_sync_and_clear_cache(self._stream)`, and the deferred, periodic, per-chunk, abort and pressure-reclaim clears all use it. — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/scheduler.py) - Scheduler call sites include a clear every 1024 decode tokens and a deferred clear after request end (`_schedule_deferred_metal_clear`), whose comment says an immediate clear after completion races. — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/scheduler.py) - Scheduler wraps its error-recovery clear in try/except because `mx.synchronize()` or `mx.clear_cache()` can raise a C++ exception that would SIGABRT (issue 435). — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/scheduler.py) ## Corrections and disagreements - CONTRADICTS: omlx-metal-cache-clears-wait-for-the-owning-stre.md open question implies the sites may have been converted; on main only the `vlm_mtp.py` and GLM decode sites are, and the engine, engine_pool, dflash, vlm and oq sites are not. — [source](https://raw.githubusercontent.com/jundot/omlx/main/omlx/engine_pool.py)