github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #3038] Log says "Nvidia GPU detected" and then "no GPU detected" #1868

New Issue
Closed
opened 2026-04-12 11:56:27 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-12 11:56:27 -05:00
Owner
Copy Link

Originally created by @jimstevens2001 on GitHub (Mar 10, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/3038

I am running a fresh install of Ollama inside of an Ubuntu 22.04 VM running an Nvidia RTX 4090 via pci passthrough (installed with "curl -fsSL https://ollama.com/install.sh | sh"). I have verified that nvidia-smi works as expected and a pytorch program can detect the GPU, but when I run Ollama, it uses the CPU to execute. Note that I have an almost identical setup (except on the host rather than in a guest) running a version of Ollama from late December with "ollama run mixtral:8x7b-instruct-v0.1-q2_K" and it uses the GPU properly.

Here is the log output that shows the inconsistent messages "Nvidia GPU detected" and then "no GPU detected"...

Mar 10 07:14:14 hinton systemd[1]: Started Ollama Service.
Mar 10 07:14:14 hinton ollama[998]: time=2024-03-10T07:14:14.622Z level=INFO source=images.go:710 msg="total blobs: 0"
Mar 10 07:14:14 hinton ollama[998]: time=2024-03-10T07:14:14.622Z level=INFO source=images.go:717 msg="total unused blobs removed: 0"
Mar 10 07:14:14 hinton ollama[998]: time=2024-03-10T07:14:14.622Z level=INFO source=routes.go:1021 msg="Listening on 127.0.0.1:11434 (version 0.1>
Mar 10 07:14:14 hinton ollama[998]: time=2024-03-10T07:14:14.622Z level=INFO source=payload_common.go:107 msg="Extracting dynamic libraries..."
Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.159Z level=INFO source=payload_common.go:146 msg="Dynamic LLM libraries [cpu_avx2 ro>
Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.159Z level=INFO source=gpu.go:94 msg="Detecting GPU type"
Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.159Z level=INFO source=gpu.go:265 msg="Searching for GPU management library libnvidi>
Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.160Z level=INFO source=gpu.go:311 msg="Discovered GPU libraries: [/usr/lib/x86_64-li>
Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.165Z level=INFO source=gpu.go:99 msg="Nvidia GPU detected"
Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.165Z level=INFO source=cpu_common.go:18 msg="CPU does not have vector extensions"
Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.165Z level=WARN source=gpu.go:128 msg="CPU does not have AVX or AVX2, disabling GPU >
Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.165Z level=INFO source=routes.go:1044 msg="no GPU detected"

Originally created by @jimstevens2001 on GitHub (Mar 10, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/3038 I am running a fresh install of Ollama inside of an Ubuntu 22.04 VM running an Nvidia RTX 4090 via pci passthrough (installed with "curl -fsSL https://ollama.com/install.sh | sh"). I have verified that nvidia-smi works as expected and a pytorch program can detect the GPU, but when I run Ollama, it uses the CPU to execute. Note that I have an almost identical setup (except on the host rather than in a guest) running a version of Ollama from late December with "ollama run mixtral:8x7b-instruct-v0.1-q2_K" and it uses the GPU properly. Here is the log output that shows the inconsistent messages "Nvidia GPU detected" and then "no GPU detected"... Mar 10 07:14:14 hinton systemd[1]: Started Ollama Service. Mar 10 07:14:14 hinton ollama[998]: time=2024-03-10T07:14:14.622Z level=INFO source=images.go:710 msg="total blobs: 0" Mar 10 07:14:14 hinton ollama[998]: time=2024-03-10T07:14:14.622Z level=INFO source=images.go:717 msg="total unused blobs removed: 0" Mar 10 07:14:14 hinton ollama[998]: time=2024-03-10T07:14:14.622Z level=INFO source=routes.go:1021 msg="Listening on 127.0.0.1:11434 (version 0.1> Mar 10 07:14:14 hinton ollama[998]: time=2024-03-10T07:14:14.622Z level=INFO source=payload_common.go:107 msg="Extracting dynamic libraries..." Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.159Z level=INFO source=payload_common.go:146 msg="Dynamic LLM libraries [cpu_avx2 ro> Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.159Z level=INFO source=gpu.go:94 msg="Detecting GPU type" Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.159Z level=INFO source=gpu.go:265 msg="Searching for GPU management library libnvidi> Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.160Z level=INFO source=gpu.go:311 msg="Discovered GPU libraries: [/usr/lib/x86_64-li> Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.165Z level=INFO source=gpu.go:99 msg="Nvidia GPU detected" Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.165Z level=INFO source=cpu_common.go:18 msg="CPU does not have vector extensions" Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.165Z level=WARN source=gpu.go:128 msg="CPU does not have AVX or AVX2, disabling GPU > Mar 10 07:14:16 hinton ollama[998]: time=2024-03-10T07:14:16.165Z level=INFO source=routes.go:1044 msg="no GPU detected"
GiteaMirror commented 2026-04-12 11:56:29 -05:00
Author
Owner
Copy Link

@jimstevens2001 commented on GitHub (Mar 10, 2024):

Found related issues...

Looks like it is currently required that the guest has AVX support to use the GPU mode.

Adding this to my qemu command line fixed it...

-cpu Haswell,+avx,+avx2 \

Note that I originally tried "-cpu host" and that did not work. It just entered the qemu monitor for some reason.

<!-- gh-comment-id:1987212816 --> @jimstevens2001 commented on GitHub (Mar 10, 2024): Found related issues... - https://github.com/ollama/ollama/issues/2524 - https://github.com/ollama/ollama/issues/2903 - https://github.com/ollama/ollama/issues/2187 Looks like it is currently required that the guest has AVX support to use the GPU mode. Adding this to my qemu command line fixed it... -cpu Haswell,+avx,+avx2 \ Note that I originally tried "-cpu host" and that did not work. It just entered the qemu monitor for some reason.
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
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#1868
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?