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 #7952] Problems (with nvidia-smi) after upgrading to 0.4.7 (from 0.3 series) #5091

New Issue
Closed
opened 2026-04-12 16:11:19 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 16:11:19 -05:00
Owner
Copy Link

Originally created by @stronk7 on GitHub (Dec 5, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/7952

What is the issue?

Hi,

while testing the new 0.4.7 series, everything seems to be working ok (Mac), but I've detected a problem when running on Ubuntu 24.04, with docker.

And, more specifically, the problem is with nvidia-smi because, unless I'm wrong, the GPU is being used normally and not the CPU.

With 0.3.13, I get this (all correct):

ollama0 3 13

But once I switch to 0.4.7, I get this:

ollama0 4 7

Note that, while the "total" memory usage displays ok (more or less similar to the 0.3.x one), the memory in the processes is not shown at all, although the processes themselves are.

I've looked to logs and everything looks ok there. And the only changing piece is the ollama docker version.

I was planning to go playing with the new Flash attention and K/V improvements and that piece of information is vital to be able to compare with different models and context sizes, so it would be great to get it back (if it's something somehow related with Ollama).

So, that's the reason for reporting it. Thanks for all the hard work, you rock!

Ciao :-)

OS

Docker

GPU

Nvidia

CPU

AMD

Ollama version

0.4.7

Originally created by @stronk7 on GitHub (Dec 5, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/7952 ### What is the issue? Hi, while testing the new 0.4.7 series, everything seems to be working ok (Mac), but I've detected a problem when running on Ubuntu 24.04, with docker. And, more specifically, the problem is with `nvidia-smi` because, unless I'm wrong, the GPU is being used normally and not the CPU. With 0.3.13, I get this (all correct): ![ollama0 3 13](https://github.com/user-attachments/assets/7ed47eea-cce0-434b-8e67-781411a2e371) But once I switch to 0.4.7, I get this: ![ollama0 4 7](https://github.com/user-attachments/assets/84dbc896-be72-47d3-8271-988ae7ac20ee) Note that, while the "total" memory usage displays ok (more or less similar to the 0.3.x one), the memory in the processes is not shown at all, although the processes themselves are. I've looked to logs and everything looks ok there. And the only changing piece is the ollama docker version. I was planning to go playing with the new Flash attention and K/V improvements and that piece of information is vital to be able to compare with different models and context sizes, so it would be great to get it back (if it's something somehow related with Ollama). So, that's the reason for reporting it. Thanks for all the hard work, you rock! Ciao :-) ### OS Docker ### GPU Nvidia ### CPU AMD ### Ollama version 0.4.7
GiteaMirror added the bug label 2026-04-12 16:11:19 -05:00
GiteaMirror commented 2026-04-12 16:11:20 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Dec 5, 2024):

I've noticed something similar when I have GGML_CUDA_ENABLE_UNIFIED_MEMORY=1 set: the amount of memory reported per runner is much lower (not 0 as in your case). Works fine tho.

<!-- gh-comment-id:2521137164 --> @rick-github commented on GitHub (Dec 5, 2024): I've noticed something similar when I have `GGML_CUDA_ENABLE_UNIFIED_MEMORY=1` set: the amount of memory reported per runner is much lower (not 0 as in your case). Works fine tho.
GiteaMirror commented 2026-04-12 16:11:21 -05:00
Author
Owner
Copy Link

@stronk7 commented on GitHub (Jan 16, 2025):

For the records, I've just updated to 0.5.6 and it seems that the problems reported here with 0.4.x are gone.

So, maybe, this can be closed now. Proceeding.

Ciao :-)

<!-- gh-comment-id:2596134572 --> @stronk7 commented on GitHub (Jan 16, 2025): For the records, I've just updated to 0.5.6 and it seems that the problems reported here with 0.4.x are gone. So, maybe, this can be closed now. Proceeding. Ciao :-)
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#5091
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?