Community

Build a better hardware record

Get help, share a reproducible result, or report a regression. Pantheon’s public channels are organized around useful, actionable evidence.

GitHub Discussions

Open Pantheon Discussions

Discussions live on the source repository, the one pip install pantheon-gpu and the README point at. Use them for installation and configuration questions, benchmark showcases, workload ideas, feature requests, and release feedback. Two threads are pinned: post your report and card, and what Pantheon should test next.

Before running a stress workload, read the workload documentation and monitor temperatures, power, and stability. Do not run hardware beyond conditions you are authorized and equipped to test.

Submit a benchmark

Submit a benchmark result

Attach the generated pantheon_report_*.json file along with the GPU, driver, OS, command, duration, and cooling/power conditions. Remove information you do not want published. Submitted results are reviewed before they are added to the dashboard.

Report a regression

Report a hardware regression

Use the hardware-regression form for a repeatable crash, display artifact, driver reset, system/WSL hang, unexpected throttling, or performance regression. Include the last known good Pantheon version and a minimal reproduction command.

Discord launch policy

Pantheon will publish a Discord invite only after two named moderators are available and the following baseline is configured:

1. Read-only #announcements and public #getting-started, #benchmark-showcase, #support, and #contributors channels.
2. Server rules that match the Code of Conduct.
3. Discord's moderation and verification controls enabled before invitations are shared.
4. A clear routing policy: reproducible bugs go to GitHub Issues; questions and results stay in Discussions or Discord.

Until then, GitHub Discussions is Pantheon's official public support channel.