[GH-ISSUE #8286] Allow use of locally installed CUDA or ROCm #51813

Open
opened 2026-04-28 20:59:28 -05:00 by GiteaMirror · 1 comment
Owner

Originally created by @erkinalp on GitHub (Jan 2, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8286

Ollama tries to install its own copy of CUDA or ROCm, even when the same version is already installed as a system-wide installation

Originally created by @erkinalp on GitHub (Jan 2, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8286 Ollama tries to install its own copy of CUDA or ROCm, even when the same version is already installed as a system-wide installation
GiteaMirror added the feature request label 2026-04-28 20:59:28 -05:00
Author
Owner

@erkinalp commented on GitHub (Jan 3, 2025):

Tested this on a ROCm setup by manually symlinking the rocblas folder to the existing ROCm installation and it works fine

<!-- gh-comment-id:2568901218 --> @erkinalp commented on GitHub (Jan 3, 2025): Tested this on a ROCm setup by manually symlinking the rocblas folder to the existing ROCm installation and it works fine
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#51813