https://github.com/kajdev/runpod-cuda-version-reporter

A runpod serverless worker that just reports the CUDA version of the machine
https://github.com/kajdev/runpod-cuda-version-reporter

Last synced: 1 day ago
JSON representation

A runpod serverless worker that just reports the CUDA version of the machine

• Host: GitHub
• URL: https://github.com/kajdev/runpod-cuda-version-reporter
• Owner: KAJdev
• License: mit
• Created: 2024-02-12T19:47:17.000Z (almost 2 years ago)
• Default Branch: main
• Last Pushed: 2025-02-27T22:21:05.000Z (10 months ago)
• Last Synced: 2025-02-28T07:14:30.351Z (10 months ago)
• Language: Dockerfile
• Size: 22.5 KB
• Stars: 0
• Watchers: 1
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# runpod-cuda-version-reporter

A tiny RunPod Serverless worker whose *entire job* is to tell you what CUDA version it's running on. That’s it. No frills, no hidden features, no actual productivity value. It’s the digital equivalent of asking, “Hey, you up?” — but for GPUs.

## Useful for:

- Debugging mysterious CUDA version mismatches  

- Verifying whether your container is actually using the base image you *think* it is  

- Poking at RunPod’s serverless runtime just to see if it responds  

## Not useful for:

- Literally anything else

If you’re expecting this to do inference, training, or anything remotely intelligent — it won’t. It prints `nvcc --version` and leaves. A hero’s work.