github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 21:25:46 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #15025] CPU bound delays #9652

New Issue
Open
opened 2026-04-12 22:32:34 -05:00 by GiteaMirror · 7 comments
GiteaMirror commented 2026-04-12 22:32:34 -05:00
Owner
Copy Link

Originally created by @v0l on GitHub (Mar 23, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/15025

What is the issue?

During long sessions in opencode, i noticed that the GPU is idle waiting for something, the GPU is idle so there is some CPU process bottleneck, notice the CPU spikes (single threaded work) between GPU spikes.

Image Image

GPU: RTX Pro 6000 MaxQ
CPU: Intel Core i9-14900k
RAM: 64GB DDR5
Model: qwen3.5:122b

Relevant log output

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.17.5

Originally created by @v0l on GitHub (Mar 23, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/15025 ### What is the issue? During long sessions in opencode, i noticed that the GPU is idle waiting for something, the GPU is idle so there is some CPU process bottleneck, notice the CPU spikes (single threaded work) between GPU spikes. <img width="1253" height="422" alt="Image" src="https://github.com/user-attachments/assets/90fed92a-25b6-40fa-b9cb-fd145d71494f" /> <img width="1258" height="481" alt="Image" src="https://github.com/user-attachments/assets/e9111484-2e42-4332-b140-d7a9e0469be8" /> GPU: RTX Pro 6000 MaxQ CPU: Intel Core i9-14900k RAM: 64GB DDR5 Model: qwen3.5:122b ### Relevant log output ```shell ``` ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.17.5
GiteaMirror added the bug label 2026-04-12 22:32:34 -05:00
GiteaMirror commented 2026-04-12 22:32:35 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 23, 2026):

What's the output of ollama ps?

<!-- gh-comment-id:4114210597 --> @rick-github commented on GitHub (Mar 23, 2026): What's the output of `ollama ps`?
GiteaMirror commented 2026-04-12 22:32:36 -05:00
Author
Owner
Copy Link

@BillionClaw commented on GitHub (Mar 24, 2026):

Looking into this — CPU-bound delays suggest a threading or processing bottleneck. I'll investigate and identify the root cause. Happy to submit a fix.

<!-- gh-comment-id:4115503639 --> @BillionClaw commented on GitHub (Mar 24, 2026): Looking into this — CPU-bound delays suggest a threading or processing bottleneck. I'll investigate and identify the root cause. Happy to submit a fix.
GiteaMirror commented 2026-04-12 22:32:36 -05:00
Author
Owner
Copy Link

@v0l commented on GitHub (Mar 24, 2026):

What's the output of ollama ps?

100% GPU

<!-- gh-comment-id:4116697004 --> @v0l commented on GitHub (Mar 24, 2026): > What's the output of `ollama ps`? 100% GPU
GiteaMirror commented 2026-04-12 22:32:37 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 24, 2026):

Server logs will aid in debugging.

<!-- gh-comment-id:4116703631 --> @rick-github commented on GitHub (Mar 24, 2026): [Server logs](https://docs.ollama.com/troubleshooting) will aid in debugging.
GiteaMirror commented 2026-04-12 22:32:37 -05:00
Author
Owner
Copy Link

@v0l commented on GitHub (Mar 24, 2026):

ollama_debug.txt

Between evaluating already loaded and completion request and loading cache slot there seems to be the CPU bound work here, can be upwards of 10s of CPU bound work.

<!-- gh-comment-id:4116789752 --> @v0l commented on GitHub (Mar 24, 2026): [ollama_debug.txt](https://github.com/user-attachments/files/26209165/ollama_debug.txt) Between `evaluating already loaded` and `completion request` and `loading cache slot` there seems to be the CPU bound work here, can be upwards of 10s of CPU bound work.
GiteaMirror commented 2026-04-12 22:32:37 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 24, 2026):

What client are you using? nmv, opencode.

<!-- gh-comment-id:4116987059 --> @rick-github commented on GitHub (Mar 24, 2026): ~What client are you using?~ nmv, opencode.
GiteaMirror commented 2026-04-12 22:32:38 -05:00
Author
Owner
Copy Link

@v0l commented on GitHub (Mar 24, 2026):

Dont seem to have the same issue running llama.cpp directly:

./build/bin/llama-server -hf unsloth/Qwen3.5-122B-A10B-GGUF:Q4_K_M -a qwen3.5:122b
<!-- gh-comment-id:4117335637 --> @v0l commented on GitHub (Mar 24, 2026): Dont seem to have the same issue running `llama.cpp` directly: ```bash ./build/bin/llama-server -hf unsloth/Qwen3.5-122B-A10B-GGUF:Q4_K_M -a qwen3.5:122b ```
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 bug
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#9652
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?