runpod-torch-v280 & RTX 4090 unsatisfied condition: cuda>=12.8

Hello, start container for runpod/pytorch:2.8.0-py3.11-cuda12.8.1-cudnn-devel-ubuntu22.04: begin error starting container: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy' nvidia-container-cli: requirement error: unsatisfied condition: cuda>=12.8, please update your driver to a newer version, or use an earlier cuda container: unknown Should I use another Template ?
5 Replies
Dj
Dj3d ago
Yes, there is a CUDA 12.8 template you can choose when you deploy your pod. If you need a link to it I can get you one.
Glee
Glee3d ago
does this template work with rtx 5090 for vllm as well?
Yiannis
YiannisOP3d ago
Tnx Dj, this is what I get error https://www.runpod.io/console/explore/runpod-torch-v280 (runpod/pytorch:2.8.0-py3.11-cuda12.8.1-cudnn-devel-ubuntu22.04), is there another ? Please send the link.
Dj
Dj3d ago
Oh sorry originally I misread the message. None of the 4090 servers have CUDA 12.8 on the server, you can get 12.8 support on the 3090 and 5090 devices!
Madiator2011
Madiator20113d ago
use filter option when deploying pod to pick only host machines with cuda version 12.8

Did you find this page helpful?