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 #11413] runner: Pass custom thread count to backend via env varibale #44781

New Issue
Open
opened 2026-04-25 00:27:10 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-25 00:27:10 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/11413
Author: @shalinib-ibm
Created: 7/14/2025
Status: 🔄 Open

Base: main ← Head: set_env_threads

📝 Commits (2)

  • bffd076 PowerPC: Pass Optimal thread count to backend
  • 529675e Merge branch 'main' into set_env_threads

📊 Changes

1 file changed (+47 additions, -1 deletions)

View changed files

📝 runner/llamarunner/runner.go (+47 -1)

📄 Description

This patch allows users to override the number of threads used during model execution via the OLLAMA_NUM_THREADS environment variable.

If this variable is not set, an optimal thread count is set for Power Arch only

By default, the thread count is set to runtime.NumCPU(), which uses all available logical cores. This can cause CPU thrashing and degraded performance.

In internal testing, tuning the thread count to an optimal value for the granite-3b:8b model led to a 40% improvement in inference throughput, compared to the default thread setting.

Example usage:
OLLAMA_NUM_THREADS=8 ollama run ...

🔄 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/11413 **Author:** [@shalinib-ibm](https://github.com/shalinib-ibm) **Created:** 7/14/2025 **Status:** 🔄 Open **Base:** `main` ← **Head:** `set_env_threads` --- ### 📝 Commits (2) - [`bffd076`](https://github.com/ollama/ollama/commit/bffd0763f7e007243d26ce6f1b656a30b3da2b68) PowerPC: Pass Optimal thread count to backend - [`529675e`](https://github.com/ollama/ollama/commit/529675e958ec037dfa4be213bdcaadc37fae56f2) Merge branch 'main' into set_env_threads ### 📊 Changes **1 file changed** (+47 additions, -1 deletions) <details> <summary>View changed files</summary> 📝 `runner/llamarunner/runner.go` (+47 -1) </details> ### 📄 Description This patch allows users to override the number of threads used during model execution via the OLLAMA_NUM_THREADS environment variable. If this variable is not set, an optimal thread count is set for Power Arch only By default, the thread count is set to runtime.NumCPU(), which uses all available logical cores. This can cause CPU thrashing and degraded performance. In internal testing, tuning the thread count to an optimal value for the `granite-3b:8b` model led to a **40% improvement in inference throughput**, compared to the default thread setting. Example usage: OLLAMA_NUM_THREADS=8 ollama run ... --- <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 2026-04-25 00:27:10 -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#44781
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?