If this comment is federating then I started hosting my first service – Lemmy itself.

Get something new enough and continue getting something new enough when AMD pushes them out. The drivers suck for anything older than an RX580, and things like Blender require even newer GPUs despite the hardware being more than capable.

Run Arch and use the ROCm’d PyTorch from the repos. Those packagers know what they’re doing.

Other than that, expect everything premade to be made for CUDA. There are some tools like https://github.com/ROCm-Developer-Tools/HIPIFY but they aren’t “there”.

Source: Been running Stable Diffusion on an RX580.

Welcome to 196. Now post before you leave.

ok this is now suddenly less fun

install the izzyondroid repo and reinstall from there.

I’m running this instance I’m posting this from (single user) on a Hetzner CX21 (2 core / 4 GB ram / 40 GB storage)

The current stats (according to podman) are as follows

NAME        CPU %       MEM USAGE / LIMIT  MEM %       NET IO             BLOCK IO           PIDS        CPU TIME       AVG CPU %
postgres    0.41%       95.12MB / 4.005GB  2.38%       1.863GB / 346.5MB  240.6MB / 10.28GB  11          16m26.041917s  0.55%
caddy       0.03%       54.47MB / 4.005GB  1.36%       244.9MB / 239.3MB  138.7MB / 20MB     9           2m41.545823s   0.40%
lemmy-be    0.10%       22.95MB / 4.005GB  0.57%       627.4MB / 2.831GB  169.4MB / 13.71MB  5           10m25.284369s  1.30%
pict-rs     0.15%       32.23MB / 4.005GB  0.80%       207.1MB / 264.7MB  584.5MB / 460.3MB  13          14m32.532772s  1.82%
lemmy-ui    0.02%       37MB / 4.005GB     0.92%       53.42MB / 10.18MB  231.1MB / 34.19MB  12          26.062038s     0.05%

Net IO is inaccurate as that also includes networking between the individual containers (I’m definitely not exposing postgres to the wider world, at least not intentionally)

Oh, and images are using about 200MB disk space so far, although this server hasn’t been up for more than 24 hours even, so we’ll see how that goes.

Is it shitty setting an instance up or is it doable?

Definitely needs some work. The instructions are all over the place, the Ansible roles expect to be the only thing in your system (unfortunately common), both the manual and Docker installation guides are meh and partially unsupported. Caddy is a config snippet and “good luck” (implied, they don’t put it that bluntly). I’m sure there are a fair few people who are accidentally running a dev-focused setup on prod.

TLDR: It’s definitely not for the newcomer. I may throw my own configs up somewhere for future reference.

Are u running on perm or via a provider?

Not sure what you mean by that.

There was an idea from blahaj.zone who considered making down votes enabled, but have less “weight” compared to upvotes. It sounded pretty interesting but I don’t think they implemented it yet.

You create an account there probably because you are a star trek fan and want to show it off

Hell, they could’ve disabled account registration and just hosted the communities. Lemmy allows for that kind of flexibility

Xorg is not even in maintenance mode at this point. It’s practically a zombie (and the devs are pretty clear that if you want that to change, YOU will have to step up to do it)

Wayland has the basics done bar Gnome being Gnome and Nvidia being Nvidia, and the uncommon use cases are having solutions built for them as we speak, although quality software will inevitably take time. Especially if we don’t want Wayland to end up an Xorg v2, but splintered.