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 #11729] GPT-OSS 120B just use CPU #7767

New Issue
Closed
opened 2026-04-12 19:55:26 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 19:55:26 -05:00
Owner
Copy Link

Originally created by @ZYJZYJZYJ0801 on GitHub (Aug 6, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/11729

What is the issue?

OS: win11
Ollama version: 0.11.2
GPU: 5000Ada*3

NAME SIZE PROCESSOR CONTEXT UNTIL
gpt-oss:120b 67 GB 100% CPU 8192 3 minutes from now

Relevant log output

OS

No response

GPU

No response

CPU

No response

Ollama version

No response

Originally created by @ZYJZYJZYJ0801 on GitHub (Aug 6, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/11729 ### What is the issue? OS: win11 Ollama version: 0.11.2 GPU: 5000Ada*3 NAME SIZE PROCESSOR CONTEXT UNTIL gpt-oss:120b 67 GB **100% CPU** 8192 3 minutes from now ### Relevant log output ```shell ``` ### OS _No response_ ### GPU _No response_ ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the bug label 2026-04-12 19:55:26 -05:00
GiteaMirror commented 2026-04-12 19:55:28 -05:00
Author
Owner
Copy Link

@ajunca commented on GitHub (Aug 6, 2025):

I had a similar problem with the 20B model. I could force setting the offloading layers to GPU on OpenWebUI to max and then the GPU is used.
Now the problem is that it crashes mid answering. I have two RTX 3090, and one always crashes mid response. Not a hardware problem, and all the other models working correctly.

I'm attaching log, but probably the most relevant lines are:

panic: failed to sample token: sample: logits sum to NaN, check model output

goroutine 12 [running]:
github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc0005bf8c0, {0x55bd8ee85af0, 0xc000383630})
	github.com/ollama/ollama/runner/ollamarunner/runner.go:364 +0x65
created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1
	github.com/ollama/ollama/runner/ollamarunner/runner.go:960 +0xa74

log.txt

<!-- gh-comment-id:3158638017 --> @ajunca commented on GitHub (Aug 6, 2025): I had a similar problem with the 20B model. I could force setting the offloading layers to GPU on OpenWebUI to max and then the GPU is used. Now the problem is that it crashes mid answering. I have two RTX 3090, and one always crashes mid response. Not a hardware problem, and all the other models working correctly. I'm attaching log, but probably the most relevant lines are: ``` panic: failed to sample token: sample: logits sum to NaN, check model output goroutine 12 [running]: github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc0005bf8c0, {0x55bd8ee85af0, 0xc000383630}) github.com/ollama/ollama/runner/ollamarunner/runner.go:364 +0x65 created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1 github.com/ollama/ollama/runner/ollamarunner/runner.go:960 +0xa74 ``` [log.txt](https://github.com/user-attachments/files/21615080/log.txt)
GiteaMirror commented 2026-04-12 19:55:28 -05:00
Author
Owner
Copy Link

@danielfl commented on GitHub (Aug 6, 2025):

@ZYJZYJZYJ0801 Are you using an interface or the direct ollama cli?
I get the same issue using openwebui - ubuntu with AMD card that fits the 20b model (not the 120b).
Using the CLI worked on the GPU during a few tests. Saying "Hi" on openwebui makes the model to move to the CPU.

https://github.com/open-webui/open-webui/issues/16303#issuecomment-3157788670

<!-- gh-comment-id:3159817397 --> @danielfl commented on GitHub (Aug 6, 2025): @ZYJZYJZYJ0801 Are you using an interface or the direct ollama cli? I get the same issue using openwebui - ubuntu with AMD card that fits the 20b model (not the 120b). Using the CLI worked on the GPU during a few tests. Saying "Hi" on openwebui makes the model to move to the CPU. https://github.com/open-webui/open-webui/issues/16303#issuecomment-3157788670
GiteaMirror commented 2026-04-12 19:55:29 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Aug 6, 2025):

There was a bug in 0.11.2 and below where the memory estimation would become too high for gpt-oss if the model needed to be split across GPU and CPU or multiple GPUs. This would often cause 100% on CPU usage once the model overflows a single GPU.

This is fixed in 0.11.3.

<!-- gh-comment-id:3161750300 --> @jessegross commented on GitHub (Aug 6, 2025): There was a bug in 0.11.2 and below where the memory estimation would become too high for gpt-oss if the model needed to be split across GPU and CPU or multiple GPUs. This would often cause 100% on CPU usage once the model overflows a single GPU. This is fixed in 0.11.3.
GiteaMirror commented 2026-04-12 19:55:29 -05:00
Author
Owner
Copy Link

@alienatedsec commented on GitHub (Aug 7, 2025):

There was a bug in 0.11.2 and below where the memory estimation would become too high for gpt-oss if the model needed to be split across GPU and CPU or multiple GPUs. This would often cause 100% on CPU usage once the model overflows a single GPU.

This is fixed in 0.11.3.

@jessegross it's much better, thanks, but please see https://github.com/ollama/ollama/issues/11676#issuecomment-3166035491

<!-- gh-comment-id:3166042276 --> @alienatedsec commented on GitHub (Aug 7, 2025): > There was a bug in 0.11.2 and below where the memory estimation would become too high for gpt-oss if the model needed to be split across GPU and CPU or multiple GPUs. This would often cause 100% on CPU usage once the model overflows a single GPU. > > This is fixed in 0.11.3. @jessegross it's much better, thanks, but please see https://github.com/ollama/ollama/issues/11676#issuecomment-3166035491
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#7767
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?