github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[PR #15845] [MERGED] Update MLX and MLX-C with threading fixes #77619

New Issue
Closed
opened 2026-05-05 10:17:33 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-05-05 10:17:33 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/15845
Author: @dhiltgen
Created: 4/27/2026
Status: ✅ Merged
Merged: 5/3/2026
Merged by: @dhiltgen

Base: main ← Head: mlx-bump-threading

📝 Commits (5)

  • 2914549 Update MLX and MLX-C
  • 8517013 Run MLX CGO work on a locked OS thread
  • 20db72d mlx: use common status writer
  • 864b2c2 mlx: bundle missing libjaccl on arm64
  • 524bc22 review comments

📊 Changes

29 files changed (+2119 additions, -276 deletions)

View changed files

📝 CMakeLists.txt (+8 -0)
📝 MLX_C_VERSION (+1 -1)
📝 MLX_VERSION (+1 -1)
📝 llm/status.go (+17 -6)
📝 llm/status_test.go (+42 -18)
📝 scripts/build_darwin.sh (+4 -2)
📝 x/imagegen/mlx/mlx.c (+339 -49)
📝 x/imagegen/mlx/mlx.h (+146 -30)
➕ x/internal/mlxthread/thread.go (+183 -0)
➕ x/internal/mlxthread/thread_affinity_test.go (+32 -0)
➕ x/internal/mlxthread/thread_test.go (+351 -0)
➕ x/internal/mlxthread/threadid_darwin_test.go (+10 -0)
➕ x/internal/mlxthread/threadid_linux_test.go (+9 -0)
📝 x/mlxrunner/client.go (+6 -62)
📝 x/mlxrunner/mlx/array_test.go (+30 -39)
📝 x/mlxrunner/mlx/generated.c (+158 -5)
📝 x/mlxrunner/mlx/generated.h (+391 -25)
📝 x/mlxrunner/mlx/include/mlx/c/compile.h (+1 -0)
📝 x/mlxrunner/mlx/include/mlx/c/distributed_group.h (+23 -9)
📝 x/mlxrunner/mlx/include/mlx/c/fft.h (+20 -0)

...and 9 more files

📄 Description

MLX now relies on OS-thread-local execution state for streams, encoders, and caches. Add an mlxthread executor backed by runtime.LockOSThread and route runner initialization, model load, inference, status memory reads, and cleanup through the worker so Go goroutine migration cannot split MLX state across native threads.

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/ollama/ollama/pull/15845 **Author:** [@dhiltgen](https://github.com/dhiltgen) **Created:** 4/27/2026 **Status:** ✅ Merged **Merged:** 5/3/2026 **Merged by:** [@dhiltgen](https://github.com/dhiltgen) **Base:** `main` ← **Head:** `mlx-bump-threading` --- ### 📝 Commits (5) - [`2914549`](https://github.com/ollama/ollama/commit/291454953decc5884d7243830ad8ded24355fc67) Update MLX and MLX-C - [`8517013`](https://github.com/ollama/ollama/commit/85170131e8a21196d21a2ea6963512faf1a4c684) Run MLX CGO work on a locked OS thread - [`20db72d`](https://github.com/ollama/ollama/commit/20db72d14247f524a0877c5796b5adf91eeccd07) mlx: use common status writer - [`864b2c2`](https://github.com/ollama/ollama/commit/864b2c2e6c625df2ce0c8e641285432a2a489c95) mlx: bundle missing libjaccl on arm64 - [`524bc22`](https://github.com/ollama/ollama/commit/524bc223767a00b065eae582068351084126767a) review comments ### 📊 Changes **29 files changed** (+2119 additions, -276 deletions) <details> <summary>View changed files</summary> 📝 `CMakeLists.txt` (+8 -0) 📝 `MLX_C_VERSION` (+1 -1) 📝 `MLX_VERSION` (+1 -1) 📝 `llm/status.go` (+17 -6) 📝 `llm/status_test.go` (+42 -18) 📝 `scripts/build_darwin.sh` (+4 -2) 📝 `x/imagegen/mlx/mlx.c` (+339 -49) 📝 `x/imagegen/mlx/mlx.h` (+146 -30) ➕ `x/internal/mlxthread/thread.go` (+183 -0) ➕ `x/internal/mlxthread/thread_affinity_test.go` (+32 -0) ➕ `x/internal/mlxthread/thread_test.go` (+351 -0) ➕ `x/internal/mlxthread/threadid_darwin_test.go` (+10 -0) ➕ `x/internal/mlxthread/threadid_linux_test.go` (+9 -0) 📝 `x/mlxrunner/client.go` (+6 -62) 📝 `x/mlxrunner/mlx/array_test.go` (+30 -39) 📝 `x/mlxrunner/mlx/generated.c` (+158 -5) 📝 `x/mlxrunner/mlx/generated.h` (+391 -25) 📝 `x/mlxrunner/mlx/include/mlx/c/compile.h` (+1 -0) 📝 `x/mlxrunner/mlx/include/mlx/c/distributed_group.h` (+23 -9) 📝 `x/mlxrunner/mlx/include/mlx/c/fft.h` (+20 -0) _...and 9 more files_ </details> ### 📄 Description MLX now relies on OS-thread-local execution state for streams, encoders, and caches. Add an mlxthread executor backed by runtime.LockOSThread and route runner initialization, model load, inference, status memory reads, and cleanup through the worker so Go goroutine migration cannot split MLX state across native threads. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-05-05 10:17:33 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#77619
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?