[GH-ISSUE #1330] Installation downloaded cuda likely unnecessarily #62729

Closed
opened 2026-05-03 10:07:52 -05:00 by GiteaMirror · 2 comments
Owner

Originally created by @folovco on GitHub (Nov 30, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1330

Originally assigned to: @dhiltgen on GitHub.

For quick testing on a CPU-only server it would seem possibly advantageous to avoid downloading the cuda drivers if the https://ollama.ai/install.sh script can determine there is no Nvidia device to use.

Installing ollama to /usr/local/bin... >>> Creating ollama user... >>> Adding current user to ollama group...
Creating ollama systemd service... >>> Enabling and starting ollama service...
Created symlink /etc/systemd/system/default.target.wants/ollama.service → /etc/systemd/system/ollama.service.
The Ollama API is now available at 0.0.0.0:11434.
Install complete. Run "ollama" from the command line.
WARNING: No NVIDIA GPU detected. Ollama will run in CPU-only mode.

Originally created by @folovco on GitHub (Nov 30, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1330 Originally assigned to: @dhiltgen on GitHub. For quick testing on a CPU-only server it would seem possibly advantageous to avoid downloading the cuda drivers if the https://ollama.ai/install.sh script can determine there is no Nvidia device to use. >>> Installing ollama to /usr/local/bin... >>> Creating ollama user... >>> Adding current user to ollama group... >>> Creating ollama systemd service... >>> Enabling and starting ollama service... Created symlink /etc/systemd/system/default.target.wants/ollama.service → /etc/systemd/system/ollama.service. >>> The Ollama API is now available at 0.0.0.0:11434. >>> Install complete. Run "ollama" from the command line. WARNING: No NVIDIA GPU detected. Ollama will run in CPU-only mode.
GiteaMirror added the nvidia label 2026-05-03 10:07:52 -05:00
Author
Owner

@folovco commented on GitHub (Nov 30, 2023):

Also for Termux witn non-root/sudo user it fails to finish installation, which I suppose would be hard to get around..?

<!-- gh-comment-id:1833891278 --> @folovco commented on GitHub (Nov 30, 2023): Also for Termux witn non-root/sudo user it fails to finish installation, which I suppose would be hard to get around..?
Author
Owner

@dhiltgen commented on GitHub (Mar 12, 2024):

The install script requires sudo to be able to register a service with systemd.

There are 2 parts to the Nvidia stack. A driver, and user-space libraries. We bundle the libraries directly into our executable, and we require the driver to be installed IF we detect an nvidia GPU. If no GPU is detected, we do not install the driver.

If you want to "roll your own" setup and not use the install script, you can download the stand-alone executable which we publish on our github release page. For example:

wget https://ollama.com/download/ollama-linux-amd64
chmod a+x ./ollama-linux-amd64
./ollama-linux-amd64 serve
<!-- gh-comment-id:1992035261 --> @dhiltgen commented on GitHub (Mar 12, 2024): The install script requires sudo to be able to register a service with systemd. There are 2 parts to the Nvidia stack. A driver, and user-space libraries. We bundle the libraries directly into our executable, and we require the driver to be installed IF we detect an nvidia GPU. If no GPU is detected, we do not install the driver. If you want to "roll your own" setup and not use the install script, you can download the stand-alone executable which we publish on our github release page. For example: ``` wget https://ollama.com/download/ollama-linux-amd64 chmod a+x ./ollama-linux-amd64 ./ollama-linux-amd64 serve ```
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#62729