github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9047] Model process terminated abnormally #67944

New Issue
Closed
opened 2026-05-04 12:06:01 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-05-04 12:06:01 -05:00
Owner
Copy Link

Originally created by @andyaccount on GitHub (Feb 12, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9047

What is the issue?

English：I used 8 Nvidia A800 80GB graphics cards to run Orama and used Orama to run Deepseek models.
Question 1: After running the deepseek-r1:671b-q4_K_M model and starting it normally, there is a high probability that after several rounds of Q&A, a portion of the answer output will terminate abnormally. Using the Nvidia smi command, it was found that all processes running on the Nvidia graphics card had disappeared, so the running process of the model automatically exited, but Olama was still running. In response to this issue, multiple attempts have been made to restart ollama and Deepseek, but there are still cases of abnormal termination after outputting some content. This question is very frustrating, but I don't understand the specific reason for this strange problem.
Question 2: In the same environment, I tried to run the deepseek-r1:70B model, and after successfully running it, I used ollama ps to check the running status and found that 86%/14% CPU/GPU. I don't understand why 8 80GB graphics cards are all vacant, so why do they prioritize using CPU resources? Is there any way to use GPU resources 100% to perform computational tasks?
Operating environment:
Operating System Version: ubuntu 22.0.4,ollama Vension：0.5.8，CUDA Vension：12.4, Driver Vension：550.144.03

Chinese：我使用了8张英伟达的A800 80GB显卡运行ollama，并且通过ollama运行deepseek模型。
问题1: 运行deepseek-r1:671b-q4_K_M的模型，正常启动后，进行几轮问答之后，很高的概率会出现回答输出一部分就异常终止了。使用nvidia-smi指令查看，发现英伟达显卡上运行的进程已经全部消失，所以模型的运行进程自动退出了，但是ollama依旧是在运行的。针对这个问题，多次尝试重启ollama和deepseek，依旧会有输出一部分内容之后异常终止的情况。这个问题很让人抓狂，但我不明白具体是什么原因导致的这个奇怪问题。
问题2: 相同的环境下，我尝试运行deepseek-r1:70B的模型，成功运行后使用ollama ps查看运行状况，发现86%/14% CPU/GPU。我不明白8张80GB显卡都是空置的，为什么会优先使用CPU资源？有没有什么办法可以100%使用GPU资源来执行运算任务？
运行环境：
操作系统版本：ubuntu 22.0.4,ollama Vension：0.5.8，CUDA Vension：12.4, Driver Vension：550.144.03

Relevant log output

OS

Linux

GPU

Nvidia

CPU

Other

Ollama version

0.5.8

Originally created by @andyaccount on GitHub (Feb 12, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9047 ### What is the issue? English：I used 8 Nvidia A800 80GB graphics cards to run Orama and used Orama to run Deepseek models. Question 1: After running the deepseek-r1:671b-q4_K_M model and starting it normally, there is a high probability that after several rounds of Q&A, a portion of the answer output will terminate abnormally. Using the Nvidia smi command, it was found that all processes running on the Nvidia graphics card had disappeared, so the running process of the model automatically exited, but Olama was still running. In response to this issue, multiple attempts have been made to restart ollama and Deepseek, but there are still cases of abnormal termination after outputting some content. This question is very frustrating, but I don't understand the specific reason for this strange problem. Question 2: In the same environment, I tried to run the deepseek-r1:70B model, and after successfully running it, I used ollama ps to check the running status and found that 86%/14% CPU/GPU. I don't understand why 8 80GB graphics cards are all vacant, so why do they prioritize using CPU resources? Is there any way to use GPU resources 100% to perform computational tasks? Operating environment: Operating System Version: ubuntu 22.0.4,ollama Vension：0.5.8，CUDA Vension：12.4, Driver Vension：550.144.03 Chinese：我使用了8张英伟达的A800 80GB显卡运行ollama，并且通过ollama运行deepseek模型。 问题1: 运行deepseek-r1:671b-q4_K_M的模型，正常启动后，进行几轮问答之后，很高的概率会出现回答输出一部分就异常终止了。使用nvidia-smi指令查看，发现英伟达显卡上运行的进程已经全部消失，所以模型的运行进程自动退出了，但是ollama依旧是在运行的。针对这个问题，多次尝试重启ollama和deepseek，依旧会有输出一部分内容之后异常终止的情况。这个问题很让人抓狂，但我不明白具体是什么原因导致的这个奇怪问题。 问题2: 相同的环境下，我尝试运行deepseek-r1:70B的模型，成功运行后使用ollama ps查看运行状况，发现86%/14% CPU/GPU。我不明白8张80GB显卡都是空置的，为什么会优先使用CPU资源？有没有什么办法可以100%使用GPU资源来执行运算任务？ 运行环境： 操作系统版本：ubuntu 22.0.4,ollama Vension：0.5.8，CUDA Vension：12.4, Driver Vension：550.144.03 ### Relevant log output ```shell ``` ### OS Linux ### GPU Nvidia ### CPU Other ### Ollama version 0.5.8
GiteaMirror added the bug label 2026-05-04 12:06:01 -05:00
GiteaMirror commented 2026-05-04 12:06:03 -05:00
Author
Owner
Copy Link

@andyaccount commented on GitHub (Feb 12, 2025):

内存大小：24*32 = 768GB

Memory module size : 24 * 32=768GB

<!-- gh-comment-id:2653911788 --> @andyaccount commented on GitHub (Feb 12, 2025): 内存大小：24*32 = 768GB Memory module size : 24 * 32=768GB
GiteaMirror commented 2026-05-04 12:06:03 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 12, 2025):

  1. https://github.com/ollama/ollama/issues/5975
  2. What size context? Server logs will aid in debugging.
<!-- gh-comment-id:2654271897 --> @rick-github commented on GitHub (Feb 12, 2025): 1. https://github.com/ollama/ollama/issues/5975 2. What size context? [Server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) will aid in debugging.
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#67944
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?