github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-22 13:42:25 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #5498] Ollama OpenAI compatibility fails on GPU? #3440

New Issue
Closed
opened 2026-04-12 14:06:23 -05:00 by GiteaMirror · 7 comments
GiteaMirror commented 2026-04-12 14:06:23 -05:00
Owner
Copy Link

Originally created by @rhastie on GitHub (Jul 5, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/5498

What is the issue?

We have seen instances where when we use the OpenAI API compatibility layer Ollama fails to utilise our NVIDIA GPU. When we re-run the test using the Ollama generate API it does use the GPU.

Is this a configuration consideration or potentially a bug.

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.1.48

Originally created by @rhastie on GitHub (Jul 5, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/5498 ### What is the issue? We have seen instances where when we use the OpenAI API compatibility layer Ollama fails to utilise our NVIDIA GPU. When we re-run the test using the Ollama generate API it does use the GPU. Is this a configuration consideration or potentially a bug. ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.1.48
GiteaMirror added the needs more infobug labels 2026-04-12 14:06:23 -05:00
GiteaMirror commented 2026-04-12 14:06:27 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Jul 5, 2024):

This would be a bug - it should use the GPU the same way. May I ask what Nvidia hardware you may be running on? Sorry about this.

<!-- gh-comment-id:2211158665 --> @jmorganca commented on GitHub (Jul 5, 2024): This would be a bug - it should use the GPU the same way. May I ask what Nvidia hardware you may be running on? Sorry about this.
GiteaMirror commented 2026-04-12 14:06:28 -05:00
Author
Owner
Copy Link

@rhastie commented on GitHub (Jul 5, 2024):

Currently we are using a L40 NVIDIA GPU with 48Gb memory. It's part of the ADA generation so pretty new. Spec is here: https://www.nvidia.com/en-gb/data-center/l40/

We have been monitoring the GPU metrics under nvidia-smi and the utilization is zero when using the OpenAI API.

<!-- gh-comment-id:2211169221 --> @rhastie commented on GitHub (Jul 5, 2024): Currently we are using a L40 NVIDIA GPU with 48Gb memory. It's part of the ADA generation so pretty new. Spec is here: https://www.nvidia.com/en-gb/data-center/l40/ We have been monitoring the GPU metrics under nvidia-smi and the utilization is zero when using the OpenAI API.
GiteaMirror commented 2026-04-12 14:06:29 -05:00
Author
Owner
Copy Link

@Moonlight1220 commented on GitHub (Jul 8, 2024):

Considering that your GPU was made with running LLM's in mind it's most likely a bug of some kind, may I ask if you are using OpenWebUI (Formerly known as OllamaWebUI)

<!-- gh-comment-id:2215199625 --> @Moonlight1220 commented on GitHub (Jul 8, 2024): Considering that your GPU was made with running LLM's in mind it's most likely a bug of some kind, may I ask if you are using OpenWebUI (Formerly known as OllamaWebUI)
GiteaMirror commented 2026-04-12 14:06:30 -05:00
Author
Owner
Copy Link

@rhastie commented on GitHub (Jul 8, 2024):

Not currently but we can easily do that if you feel it would help.

Currently we are directly monitoring the GPU metrics and trying the respective APIs directly. At its simplest we are using "curl" to hit the two different API versions in a loop.

<!-- gh-comment-id:2215375278 --> @rhastie commented on GitHub (Jul 8, 2024): Not currently but we can easily do that if you feel it would help. Currently we are directly monitoring the GPU metrics and trying the respective APIs directly. At its simplest we are using "curl" to hit the two different API versions in a loop.
GiteaMirror commented 2026-04-12 14:06:31 -05:00
Author
Owner
Copy Link

@Moonlight1220 commented on GitHub (Jul 9, 2024):

Unsure if it will help becouse i have never seen this issue before however its worth a try, are you running Ollama localy or on a server?

<!-- gh-comment-id:2218800270 --> @Moonlight1220 commented on GitHub (Jul 9, 2024): Unsure if it will help becouse i have never seen this issue before however its worth a try, are you running Ollama localy or on a server?
GiteaMirror commented 2026-04-12 14:06:33 -05:00
Author
Owner
Copy Link

@royjhan commented on GitHub (Jul 31, 2024):

@rhastie are you still experiencing this issue?

<!-- gh-comment-id:2261050240 --> @royjhan commented on GitHub (Jul 31, 2024): @rhastie are you still experiencing this issue?
GiteaMirror commented 2026-04-12 14:06:34 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Oct 24, 2024):

If you're still seeing this, please upgrade to the latest version and if that doesn't clear it up, please share a server log and I'll reopen so we can further investigate.

<!-- gh-comment-id:2434111913 --> @dhiltgen commented on GitHub (Oct 24, 2024): If you're still seeing this, please upgrade to the latest version and if that doesn't clear it up, please share a server log and I'll reopen so we can further investigate.
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#3440
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?