github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-05 23:53:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #13152] GPT-OSS: 120B doesnt share between CPU/GPU @ CTX over 8192 #8698

New Issue
Open
opened 2026-04-12 21:28:20 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-12 21:28:20 -05:00
Owner
Copy Link

Originally created by @Stef1519 on GitHub (Nov 19, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/13152

What is the issue?

OS: Ubuntu
Ollama Version: 0.12.11
Hardware: Xeon E5 2697 V2, 2 CPUs, 128 GB RAM, 3xP106-090, 1xP106-100, Tesla M10 32 GB, Total 56 GB GPU Memory

CLI & OpenWebUI

Issue: #1

GPT-OSS:120B happily shares between CPU/GPU when using a CTX Size of 8192 (altrough nearly 50:50 wouldn't be very valid either, considering 56 GB GPU memory):

num_ctx = 8192:
Gpt-oss:120b a951a23b46a1 67 GB 48%/52% CPU/GPU 8192 4 minutes from now
Works somehow, responses not very fast, but within one Minute.

num_ctx=32768:
gpt-oss:120b a951a23b46a1 66 GB 100% CPU 32768 4 minutes from now

num_ctx=32768 and num_gpu=16:
gpt-oss:120b a951a23b46a1 68 GB 56%/44% CPU/GPU 32768 4 minutes from now
No response for ages (probably not at all, was giving up after ten minutes)

Issue #2:

GPT-OSS:20B: Takes ages to initially load, easily five minutes or more (occupying 14 GB GPU memory)

Any clues?
Thank you very much!

Relevant log output

OS

No response

GPU

No response

CPU

No response

Ollama version

No response

Originally created by @Stef1519 on GitHub (Nov 19, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/13152 ### What is the issue? OS: Ubuntu Ollama Version: 0.12.11 Hardware: Xeon E5 2697 V2, 2 CPUs, 128 GB RAM, 3xP106-090, 1xP106-100, Tesla M10 32 GB, Total 56 GB GPU Memory CLI & OpenWebUI Issue: #1 GPT-OSS:120B happily shares between CPU/GPU when using a CTX Size of 8192 (altrough nearly 50:50 wouldn't be very valid either, considering 56 GB GPU memory): num_ctx = 8192: Gpt-oss:120b a951a23b46a1 67 GB 48%/52% CPU/GPU 8192 4 minutes from now Works somehow, responses not very fast, but within one Minute. num_ctx=32768: gpt-oss:120b a951a23b46a1 66 GB 100% CPU 32768 4 minutes from now num_ctx=32768 and num_gpu=16: gpt-oss:120b a951a23b46a1 68 GB 56%/44% CPU/GPU 32768 4 minutes from now No response for ages (probably not at all, was giving up after ten minutes) Issue #2: GPT-OSS:20B: Takes ages to initially load, easily five minutes or more (occupying 14 GB GPU memory) Any clues? Thank you very much! ### Relevant log output ```shell ``` ### OS _No response_ ### GPU _No response_ ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the needs more infobug labels 2026-04-12 21:28:20 -05:00
GiteaMirror commented 2026-04-12 21:28:21 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Nov 19, 2025):

Server log will show details of layer allocation.

<!-- gh-comment-id:3553630775 --> @rick-github commented on GitHub (Nov 19, 2025): [Server log](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.mdx) will show details of layer allocation.
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 needs more info
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#8698
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?