github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-23 14:13:13 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #8872] After a period of conversation with DeepSeek R1, the memory of the server keeps decreasing and never release #52261

New Issue
Closed
opened 2026-04-28 22:41:58 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-28 22:41:58 -05:00
Owner
Copy Link

Originally created by @gallery2016 on GitHub (Feb 6, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8872

What is the issue?

Thank you for your attention.

Below is my detailed environment and configuration

Ollama client version is 0.5.7

CentOS Linux release 7.9.2009 (Core)

The GPU is: 8 * L40

Driver Version: 550.54.15 CUDA Version: 12.4

Configuration of ollam startup：
export OLLAMA_LOAD_TIMEOUT=90m
export GGML_CUDA_ENABLE_UNIFIED_MEMORY=1
export OLLAMA_GPU_OVERHEAD=536870912
export OLLAMA_FLASH_ATTENTION=1

ollama serve
ollama run deepseek-r1:671b

The total GPU memory is 48*8G=384G, if use OLLAMA to run the Q4 model of 671B, and there is not enough GPU memory, then set GGML_CUDA_ENABLE_UNIFIED_MEMORY=1, it will utilize the server's memory, and everything will be fine at startup.

However, after a period of conversation, I found that the memory of the server keeps decreasing until it runs out, and the last conversation is very laggy.

Relevant log output

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.5.7

Originally created by @gallery2016 on GitHub (Feb 6, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8872 ### What is the issue? Thank you for your attention. Below is my detailed environment and configuration Ollama client version is 0.5.7 CentOS Linux release 7.9.2009 (Core) The GPU is: 8 * L40 Driver Version: 550.54.15 CUDA Version: 12.4 Configuration of ollam startup： export OLLAMA_LOAD_TIMEOUT=90m export GGML_CUDA_ENABLE_UNIFIED_MEMORY=1 export OLLAMA_GPU_OVERHEAD=536870912 export OLLAMA_FLASH_ATTENTION=1 ollama serve ollama run deepseek-r1:671b The total GPU memory is 48*8G=384G, if use OLLAMA to run the Q4 model of 671B, and there is not enough GPU memory, then set GGML_CUDA_ENABLE_UNIFIED_MEMORY=1, it will utilize the server's memory, and everything will be fine at startup. However, after a period of conversation, I found that the memory of the server keeps decreasing until it runs out, and the last conversation is very laggy. ### Relevant log output ```shell ``` ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.5.7
GiteaMirror added the bugneeds more info labels 2026-04-28 22:42:06 -05:00
GiteaMirror commented 2026-04-28 22:42:15 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 6, 2025):

Server logs may aid in debugging.

What do you mean by runs out? Does the runner crash? Does your system fill up the swap device?

<!-- gh-comment-id:2639392287 --> @rick-github commented on GitHub (Feb 6, 2025): [Server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) may aid in debugging. What do you mean by `runs out`? Does the runner crash? Does your system fill up the swap device?
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#52261
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?