eskuero

Better than anything. I run through vulkan on lm studio because rocm on my rx 5600xt is a heavy pain

Ollama has had for a while an issue opened abou the vulkan backend but sadly it doesn't seem to be going anywhere.

Put up some docker stats in a reply in case you missed it on the refresh

For the whole stack in the past 16 hours

# docker-compose stats

Depends on how many communities do you subscribe too and how much activity they have.

I'm running my single user instance subscribed to 20 communities on a 2c/4g vps who also hosts my matrix server and a bunch of other stuff and right now I mostly see peaks from 5/10% of CPU and RAM at 1.5GB

I have been running for 15months and the docker volumes total 1.2GBs A single pg_dump for the lemmy database in plain text is 450M

Yep I also run wildcard domains for simplicity

I do the dns challenge with letsencrypt too but to not leak local dns names into the public I just run a pihole locally that can resolve those domains

What is your budget?

i5-7200U

That is Kaby Lake and seems to support up to HEVC for decoding which might be enough for you

https://en.wikipedia.org/wiki/Intel_Quick_Sync_Video#Hardware_decoding_and_encoding

For a bit of future proof you might want to check out something Tiger Lake or newer since it seems like they support AV1 decoding in hardware.

you could say it was very basic AI

Arch Linu, Kubuntu, Supergrub, Tails, Kali, Windows