What's in this build
This is a pre-release to test the per-stream-lock fix for Metal GPU crashes on macOS Tahoe (#300, #173).
Changes since v0.2.18
- Build: bundle per-stream-lock patched libmlx.dylib — Replaces stock MLX's libmlx.dylib with a patched version that adds per-stream mutex protection to Metal command buffer/encoder access. This addresses the
completeMemory() prepare count underflowkernel panic on M4 and SIGSEGV/SIGABRT crashes on M3 Ultra caused by unsynchronized concurrent GPU stream access. Based on ml-explore/mlx#3247 by @rsnow. (#300, #173) - Fix: remove keepalive warmup loop that caused sustained GPU spike (#292)
Testing needed
If you were experiencing kernel panics or server crashes on Tahoe, please try this build and report back. SSD cache enabled + heavy workload (e.g. Claude Code with tool calling) is the scenario most likely to trigger the original bug.