Maybe don't allow autonomous cars on public streets then? The tech is nowhere near ready for prime time.
jon
joined 1 year ago
I'm one of the other Lemmy.tf admins and I'll share a bit. We're currently on the docker-compose deployment from the repo, running on a VM with 4c/8gb ram/256gb disk. It's on a baremetal VMware box at OVH with loads of resources to expand as needed.
I'm hoping we get enough users on here to force me into converting to a Helm chart and moving this to my Kubernetes cluster. Pod scaling would help address some of the issues larger instances are starting to run into, and it seems like a fun project.
As for Unraid, your best bet is to see if you can install docker-compose on it. This thread from 2020 suggests it should be possible, but the binary may not persist restarts. If you can't use compose you would probably have to strip it apart and deploy one container at a time, and potentially work around the need for the Docker networks.
I may be interested in helping with an Unraid deployment guide if there's heavy interest- I'm running it on my NAS at home and can tinker a bit. Feel free to DM me if you've got questions or need any assistance.
Edit: That Unraid forum post has a reply about using a bash alias to run docker-compose in Docker, this is the route I'd go rather than having to do jank stuff to make the binary persistent. Should be able to follow the normal docker-compose install from your root user once you have compose ready. Make sure to do your port forwarding or use Nginx Proxy Manager since SSL is mandatory to federate.
AMD has ROCm which tries to get close. I've been able to get some CUDA applications running on a 6700xt, although they are noticeably slower than running on a comparable NVidia card. Maybe we'll see more projects adding native ROCm support now that AMD is trying to cater to the enterprise market.