ffhein

Price is comparable to a used RTX3090 with 24GB vram, which is probably more attractive to someone who is also interested in Linux/Windows gaming (and already owns a pc I mean). I would also guess that the RTX would be faster than the MacBook. IMO unified ram is more interesting when you can get a lot of it

2010 is ancient technology, according to wikipedia Nvidia released the 600 series in 2012… Even if there was some inference engine supporting it then lack of computational speed and memory bandwidth would probably make it not worth the effort.

I bought a used 3090 two years ago, and back then they were usually listed for €800-1000 in my country. I thought I was lucky to find one for €700 after searching for a few months, and I don’t think they’ve ever been cheaper than this here. There are definitely fewer of them available now, but you can still buy one for €950 (and possibly even lower if you’re patient). So prices have gone up, but IMO not by ridiculous amounts like RAM.

The large free search engines have really gone down the drain recently… Kagi (a paid search engine) also has a small web feature, but it’s really cool that you’re building something that isn’t profit driven. I’ll be sure to share your search engine with my friends!

Now is a bad time to buy hard drives price-wise

It’s a bit of a gamble, isn’t it? At least here, HDDs appear to have gone up 10-20% compared to lowest prices last year, which isn’t that much compared to SSDs and RAM. Personally I bought new disks last week just in case the prices continue to rise and I don’t want to end up in a position where I have to buy new disks while they’re at an all time high.

You mean “spinning disk” metaphorically, right? Or is there any reason to not have it in low power standby mode? I don’t have any hot spare in my server, but on my desktop I use hdparm to spin down a rarely used storage drive just because it’s so loud.

Never used MS Planner, but the Kanban plugin does everything I need and I use it for my larger hobby programming projects. I was already using obsidian.md for other things so for me it was very convenient.

If you don’t find any hostable service, perhaps you could try Obsidian if its Kanban plugin works well in the mobile client. It’s closed source, but all data is stored in markdown files, and you could use a self-hosted git server for storage and synchronization between users.

https://github.com/resemble-ai/chatterbox is pretty good, and has both TTS and voice cloning. Main disadvantage for me was that even if the cloning gives a consistent voice, the generated samples can get random accents.

https://huggingface.co/zai-org/GLM-TTS also seemed pretty promising, but I haven’t had time to test it yet.

I just wanted to test if it was viable to run larger MoE LLMs on CPU, e.g. Qwen3-next-80B-A3B… Even if I got acceptable generation speeds I’d probably get bored with it after a few hours, as with other local models. Had I got it for €700 it was pretty low value for money anyway, since my current RAM is enough for everything else I use the computer for. On the positive side, I can put that money towards a Steam Frame instead.

… I was thinking about buying a 96GB DDR5 kit from the local computer store a few weeks ago, but wasn’t sure it was actually worth €700. Checked again now and the exact same product costs €1500. I guess that settles it, 32GB will have to be enough for the next couple of years then.

Still are, but I guess a lot of people don’t know much about them

So far that has never happened because I’m not using that much storage :) But I shut it down when I need to turn off the mains electricity, and for powering it on afterwards the fake wall can be lifted off. It’s just the area underneath the desk so the panel might be smaller than it sounds like, and it hangs on some hooks so it’s fairly easy to remove if you know what you’re doing. Painted in the same colour as the wall, and with some some random junk on the floor in front, it blends in quite well though. I think the risk of burglary is fairly low, so it’s primarily to soothe my own paranoia.

I mounted mine on the wall under a desk in a room with no other electronics, and then put up a fake wall in front of the server. It can draw in air from the sides, and exhaust upwards behind the desk. But the only real solution is offsite backup, which will also protect against fire and other disasters.

Can’t help but think about this old XKCD from 2010.

Products targeted towards businesses have always been unreasonably more expensive than those targeted towards consumers. It sucks for us AI hobbyists that Nvidia are stingy with VRAM on consumer cards, but I don’t find it surprising.

Personally I only have a single RTX 3090, but I know a lot of people online who are stacking multiple consumer cards to run AI. Buying used 3090s and putting them in a mining rig is probably still the best value for money if you need a large amount of VRAM.

How much VRAM do you actually need btw?

Intel NUC running Linux. Not the cheapest solution but can play anything and I have full control over it. At first I tried to find some kind of programmable remote but now we have a wireless keyboard with built-in touchpad.

Biggest downside is that the hardware quality is kind of questionable and the first two broke after 3 years + a few months, so we’re on our third now.

This is my wireguard docker setup:

version: "3.6"
services:
  wireguard:
    image: linuxserver/wireguard
    container_name: wireguard
    cap_add:
      - NET_ADMIN
      - SYS_MODULE
    environment:
      - PUID=116
      - PGID=122
      - TZ=Europe/Stockholm
      - ALLOWEDIPS=192.168.1.0/24
    volumes:
      - /data/torrent/wireguard/config:/config
      - /lib/modules:/lib/modules
    ports:
      - 192.168.1.111:8122:8122  # Deluge webui
      - 192.168.1.111:9127:9127  # jackett webui
      - 192.168.1.111:9666:9666  # prowlarr webui
      - 51820:51820/udp           # wireguard
      - 192.168.1.111:58426:58426  # Deluge RPC
    sysctls:
      - net.ipv4.conf.all.src_valid_mark=1
      - net.ipv6.conf.all.disable_ipv6=1
      - net.ipv6.conf.default.disable_ipv6=1
    restart: unless-stopped

Can reach the webuis from LAN, no other network configuration was necessary. 192.168.1.111 is the server’s LAN address. The other services are configured very similar to your qbittorrent, and don’t expose any ports. Can’t promise it’s 100% correct but it’s working for me.