github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #10703] Wrong model size lead can not use GPU #7033

New Issue
Closed
opened 2026-04-12 18:56:21 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-12 18:56:21 -05:00
Owner
Copy Link

Originally created by @zhfg on GitHub (May 14, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/10703

What is the issue?

Issue description:
I use a AMD MI50 to run qwen3:32b, and open-webui as a frontend. with a same model, ollama recognition the wrong size of model in different time, it is leading can not use GPU when the wrong size.

Some output:

#Wrong size, in this time, ollama using CPU with wrong mode size
jacob@crawl:~$ ollama ps
NAME ID SIZE PROCESSOR UNTIL
qwen3:32b e1c9f234c6eb 53 GB 100% CPU 23 hours from now

#Restart ollama service and run agin, it's used GPU and right size of model
jacob@crawl:~$ ollama ps
NAME ID SIZE PROCESSOR UNTIL
qwen3:32b e1c9f234c6eb 25 GB 100% GPU 24 hours from now

Relevant log output

OS

Linux

GPU

AMD

CPU

AMD

Ollama version

0.6.8

Originally created by @zhfg on GitHub (May 14, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/10703 ### What is the issue? Issue description: I use a AMD MI50 to run qwen3:32b, and open-webui as a frontend. with a same model, ollama recognition the wrong size of model in different time, it is leading can not use GPU when the wrong size. Some output: \#Wrong size, in this time, ollama using CPU with wrong mode size jacob@crawl:~$ ollama ps NAME ID SIZE PROCESSOR UNTIL qwen3:32b e1c9f234c6eb **53 GB** 100% CPU 23 hours from now \#Restart ollama service and run agin, it's used GPU and right size of model jacob@crawl:~$ ollama ps NAME ID SIZE PROCESSOR UNTIL qwen3:32b e1c9f234c6eb **25 GB** 100% GPU 24 hours from now ### Relevant log output ```shell ``` ### OS Linux ### GPU AMD ### CPU AMD ### Ollama version 0.6.8
GiteaMirror added the bug label 2026-04-12 18:56:21 -05:00
GiteaMirror commented 2026-04-12 18:56:23 -05:00
Author
Owner
Copy Link

@mitchelldehaven commented on GitHub (May 14, 2025):

Running into a similar issue with an RTX 6000 Blackwell and qwen3:32b-q8_0

<!-- gh-comment-id:2878525851 --> @mitchelldehaven commented on GitHub (May 14, 2025): Running into a similar issue with an RTX 6000 Blackwell and `qwen3:32b-q8_0`
GiteaMirror commented 2026-04-12 18:56:24 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (May 14, 2025):

You have a client setting num_ctx to something around 124000, which requires a context buffer of 30G. This does not fit in the available VRAM with enough space for any layers, so the model is run in system RAM.

<!-- gh-comment-id:2879683633 --> @rick-github commented on GitHub (May 14, 2025): You have a client setting `num_ctx` to something around 124000, which requires a context buffer of 30G. This does not fit in the available VRAM with enough space for any layers, so the model is run in system RAM.
GiteaMirror commented 2026-04-12 18:56:24 -05:00
Author
Owner
Copy Link

@zhfg commented on GitHub (May 15, 2025):

@rick-github As my checked, the num_ctx(called Context Length (Ollama) in open webui) never changed and default is 2048.

In any case, I will keep watching on this issue and feedback if any update

<!-- gh-comment-id:2881885397 --> @zhfg commented on GitHub (May 15, 2025): @rick-github As my checked, the num_ctx(called Context Length (Ollama) in open webui) never changed and default is 2048. In any case, I will keep watching on this issue and feedback if any update
GiteaMirror commented 2026-04-12 18:56:25 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (May 15, 2025):

Server logs will aid in debugging.

<!-- gh-comment-id:2881889290 --> @rick-github commented on GitHub (May 15, 2025): [Server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) will aid in debugging.
GiteaMirror commented 2026-04-12 18:56:26 -05:00
Author
Owner
Copy Link

@zhfg commented on GitHub (May 16, 2025):

@rick-github thank you for your reply
I manually set the value of num_ctx to different values, and when I use a large value, this issue does occur.

It looks like my configuration problem. This bug is closed

<!-- gh-comment-id:2886261654 --> @zhfg commented on GitHub (May 16, 2025): @rick-github thank you for your reply I manually set the value of num_ctx to different values, and when I use a large value, this issue does occur. It looks like my configuration problem. This bug is closed
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#7033
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?