github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 13:19:40 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #9472] GPU Memory not utilized well #68228

New Issue
Closed
opened 2026-05-04 12:56:42 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-05-04 12:56:42 -05:00
Owner
Copy Link

Originally created by @forReason on GitHub (Mar 3, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9472

What is the issue?

Ollama leaves over 30% of my GPU's Memory unallocated:

+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.120                Driver Version: 550.120        CUDA Version: 12.4     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 3060        On  |   00000000:01:00.0 Off |                  N/A |
|  0%   60C    P2             88W /  100W |    7400MiB /  12288MiB |     98%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
|   1  NVIDIA RTX A2000 12GB          On  |   00000000:05:00.0 Off |                  Off |
| 30%   34C    P2             22W /   70W |    8072MiB /  12282MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
|   2  NVIDIA GeForce RTX 3060        On  |   00000000:0A:00.0 Off |                  N/A |
|  0%   50C    P2             42W /  100W |    7784MiB /  12288MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

Any Idea why that Is or what I can do to improve this?

Relevant log output

OS

Linux

GPU

Nvidia

CPU

AMD

Ollama version

0.5.4

Originally created by @forReason on GitHub (Mar 3, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9472 ### What is the issue? Ollama leaves over 30% of my GPU's Memory unallocated: ``` +-----------------------------------------------------------------------------------------+ | NVIDIA-SMI 550.120 Driver Version: 550.120 CUDA Version: 12.4 | |-----------------------------------------+------------------------+----------------------+ | GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+========================+======================| | 0 NVIDIA GeForce RTX 3060 On | 00000000:01:00.0 Off | N/A | | 0% 60C P2 88W / 100W | 7400MiB / 12288MiB | 98% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ | 1 NVIDIA RTX A2000 12GB On | 00000000:05:00.0 Off | Off | | 30% 34C P2 22W / 70W | 8072MiB / 12282MiB | 0% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ | 2 NVIDIA GeForce RTX 3060 On | 00000000:0A:00.0 Off | N/A | | 0% 50C P2 42W / 100W | 7784MiB / 12288MiB | 0% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ ``` Any Idea why that Is or what I can do to improve this? ### Relevant log output ```shell ``` ### OS Linux ### GPU Nvidia ### CPU AMD ### Ollama version 0.5.4
GiteaMirror added the bug label 2026-05-04 12:56:42 -05:00
GiteaMirror commented 2026-05-04 12:56:48 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 3, 2025):

The memory estimation is sometimes inaccurate. You can make ollama load more layers into VRAM by setting num_gpu either in an API call or in the Modelfile, see here. Note that if you set it too high, you may experience OOMs or performance degradation.

<!-- gh-comment-id:2693565629 --> @rick-github commented on GitHub (Mar 3, 2025): The memory estimation is sometimes inaccurate. You can make ollama load more layers into VRAM by setting `num_gpu` either in an API call or in the Modelfile, see [here](https://github.com/ollama/ollama/issues/6950#issuecomment-2373663650). Note that if you set it too high, you may experience OOMs or performance degradation.
GiteaMirror commented 2026-05-04 12:56:49 -05:00
Author
Owner
Copy Link

@forReason commented on GitHub (Mar 8, 2025):

this did not work:

Image

ollama ps
NAME                          ID              SIZE     PROCESSOR          UNTIL               
llama3.3:70b-instruct-q2_K    a6f03da15cbc    41 GB    12%/88% CPU/GPU    59 minutes from now

+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.120                Driver Version: 550.120        CUDA Version: 12.4     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 3060        On  |   00000000:01:00.0 Off |                  N/A |
| 61%   57C    P2             41W /  100W |   10416MiB /  12288MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
|   1  NVIDIA RTX A2000 12GB          On  |   00000000:05:00.0 Off |                  Off |
| 30%   50C    P2             28W /   70W |   10810MiB /  12282MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
|   2  NVIDIA GeForce RTX 3060        On  |   00000000:0A:00.0 Off |                  N/A |
| 33%   52C    P2             42W /  100W |   10338MiB /  12288MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
<!-- gh-comment-id:2708215988 --> @forReason commented on GitHub (Mar 8, 2025): this did not work: ![Image](https://github.com/user-attachments/assets/eb45958e-c04d-4d0f-a9c8-d36dedc781cf) ``` ollama ps NAME ID SIZE PROCESSOR UNTIL llama3.3:70b-instruct-q2_K a6f03da15cbc 41 GB 12%/88% CPU/GPU 59 minutes from now +-----------------------------------------------------------------------------------------+ | NVIDIA-SMI 550.120 Driver Version: 550.120 CUDA Version: 12.4 | |-----------------------------------------+------------------------+----------------------+ | GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+========================+======================| | 0 NVIDIA GeForce RTX 3060 On | 00000000:01:00.0 Off | N/A | | 61% 57C P2 41W / 100W | 10416MiB / 12288MiB | 0% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ | 1 NVIDIA RTX A2000 12GB On | 00000000:05:00.0 Off | Off | | 30% 50C P2 28W / 70W | 10810MiB / 12282MiB | 0% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ | 2 NVIDIA GeForce RTX 3060 On | 00000000:0A:00.0 Off | N/A | | 33% 52C P2 42W / 100W | 10338MiB / 12288MiB | 0% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ ```
GiteaMirror commented 2026-05-04 12:56:51 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 8, 2025):

The output of ollama ps will be incorrect because you have overridden num_gpu. Check the server log for the layer allocation.

<!-- gh-comment-id:2708219780 --> @rick-github commented on GitHub (Mar 8, 2025): The output of `ollama ps` will be incorrect because you have overridden `num_gpu`. Check the server log for the layer allocation.
GiteaMirror commented 2026-05-04 12:56:52 -05:00
Author
Owner
Copy Link

@forReason commented on GitHub (Mar 8, 2025):

YES!
Thank you. This may be a visual bug which could be looked into the future...

Mar 08 12:28:51 drg-ai001 ollama[4008342]: llm_load_tensors: offloaded 81/81 layers to GPU
<!-- gh-comment-id:2708222376 --> @forReason commented on GitHub (Mar 8, 2025): YES! Thank you. This may be a visual bug which could be looked into the future... ``` Mar 08 12:28:51 drg-ai001 ollama[4008342]: llm_load_tensors: offloaded 81/81 layers to GPU ```
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#68228
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?