this post was submitted on 03 Jun 2026
955 points (99.7% liked)

People Twitter

10058 readers
900 users here now

People tweeting stuff. We allow tweets from anyone.

RULES:

  1. Mark NSFW content.
  2. No doxxing people.
  3. Must be a pic of the tweet or similar. No direct links to the tweet.
  4. No bullying or international politcs
  5. Be excellent to each other.
  6. Provide an archived link to the tweet (or similar) being shown if it's a major figure or a politician. Archive.is the best way.

founded 3 years ago
MODERATORS
955
Managers (media.piefed.zip)
submitted 1 week ago* (last edited 1 week ago) by inari@piefed.zip to c/whitepeopletwitter@sh.itjust.works
 
you are viewing a single comment's thread
view the rest of the comments
[–] lime@feddit.nu 3 points 1 week ago (2 children)

i'm running moderately quantized models on 24GB VRAM and getting like 30-40 tokens a second. add a zero to the price and it's still not a lot for a company.

[–] theunknownmuncher@lemmy.world 3 points 1 week ago* (last edited 1 week ago) (1 children)

Sure, but you're running a very small model compared to what we are talking about.

GLM-5.1 is over 200GB even when quantizied to 1-bit. Kimi K2.6 is even bigger. A framework desktop cannot run either of these. Qwen3.6 is significantly smaller and the model weights could fit, but consider the KV-cache you'd need for all of the company's users, and the throughput required to serve them all.

You're right that it is within reach for a company but framework desktop makes zero sense for this

[–] lime@feddit.nu 2 points 1 week ago* (last edited 1 week ago) (3 children)

isn't qwen like 40-50GB? that could work i think. performance is okay even quantised down to 10.

[–] Evotech@lemmy.world 1 points 1 week ago (2 children)

And then add 200k context on top

And then add hundred of users needing to do things in paralell

[–] lime@feddit.nu 1 points 1 week ago

nobody said anything about it being a large company :P

anyway, seems the framework is hampered by a slow gpu so the memory issues are apparently moot.

[–] boonhet@sopuli.xyz 1 points 1 week ago (1 children)

If it's a large enough company to have hundreds of users, it can afford several beefy machines tbh

[–] Evotech@lemmy.world 1 points 1 week ago (1 children)

It's a capex and that type of hardware needs to be replaced every 3 years minimum and you need people to set it up and maintain a cluster. And it's not straight forward.

You are never going to get that approved without a serious business case.

Claude on the other end is a opex and much easier to just try out and then build a solution on it

Not saying it doesn't happen but it's not as easy as people make it sound like

[–] boonhet@sopuli.xyz 1 points 6 days ago (1 children)

It's 3 years if you're trying to be competitive on frontier models and generally capex is preferred to opex because opex never ends

I don't think anyone's building a cluster for their business right now, but one single rack after Claude gets rid of their subscription options? Might be a good deal.

[–] Evotech@lemmy.world 1 points 6 days ago (1 children)

Capex never ends either if it's hardware. Also you need opex to run it

[–] boonhet@sopuli.xyz 1 points 6 days ago (1 children)

400k on a DGX node starts seeming like a great deal when your employees each start using a few hundred dollars worth of Claude tokens every month. That one node can handle a lot of users depending on the model used.

It's an expense once every maybe 5 or 6 years in reality and you don't need to hire new people, you just give your existing sysadmins some extra work. They'll complain, but they'll still do it.

Of course the sensible alternative is to use a decent model off openrouter for peanuts but then you're sending all your sensitive business secrets to China which is even worse than sharing them with a US AI company. And people WILL be sharing secrets lol

[–] Evotech@lemmy.world 1 points 6 days ago (1 children)
[–] boonhet@sopuli.xyz 1 points 5 days ago (1 children)

You don't have to run Claude Opus for it to be useful lol

[–] Evotech@lemmy.world 1 points 5 days ago

It's always going to be second rate. And you'll have to defend that

[–] Jiral@lemmy.org 0 points 1 week ago* (last edited 1 week ago) (1 children)

At Q8 it is around 35-40GB I think + memory for required context.

I have a Framework desktop. It gets you you around 6t/s. Not suitable for professional use but for personal use I think it is fine. I do prefer Gemma 4 though, but that comes with similar reqirements.

[–] lime@feddit.nu 0 points 1 week ago (1 children)

huh, i thought that ryzen ai thing would perform better than that. my 7900xtx regularly gets 30+tps with qwen, up to hundreds with more compressed models.

[–] Jiral@lemmy.org 2 points 1 week ago* (last edited 1 week ago) (1 children)

My system runs at 100W TDP though. That is maybe 140W at the power outlet, incl. monitor and everything.

This is also the dense 27B model at Q8. But yeah, it is not terribly fast. I think the best use case is on MoE models. GPT-OSS-120B runs on it for example and at 50T/s speed is not a n issue anymore either. (I could get it to run even on just 64GB but the new llama.cpp might need a tiny bit more memory which pushed it just across the limit. yeah I know, for seriously using it you'd need the 128GB version)

[–] lime@feddit.nu 2 points 1 week ago (1 children)

that's fair, i'm at like 7x the power. the gpu alone easily pulls 350-400W and the rest of the system isn't exactly running lean either.

...man now i really want more vram.

[–] Jiral@lemmy.org 1 points 1 week ago* (last edited 1 week ago) (1 children)

Yes I think Strix Halo makes sense when low power use is a requirement. I built a custom fanless Strix Halo system for the fun of it and I guess there aren't too many out there running Gemma 4 31B Q8 without a single fan, anywhere.

And for MoE models that need 60-80GB + context it is perfect. Those are decently fast then as well.

PS: If VRAM is all you care about the maxed out Mac Studio is fascinating. 512GB unified memory for around 10K EUR (pre crazy bubble prices) That should be able to run pretty large MoE models but dense models of that size would probably run glacially.

[–] lime@feddit.nu 1 points 1 week ago (1 children)

i'm not buying any hardware for the forseeable future :P it's all just wishful thinking at the moment. but unified memory architectureis probably going to become more common so maybe in five years when some new motherboard standard becomes the norm...

[–] Jiral@lemmy.org 1 points 1 week ago* (last edited 1 week ago)

I fully understand. ;) Buying hardware now means you'd be either crazy or desperate.

[–] Evotech@lemmy.world 1 points 1 week ago

Yes. It will probably work for 1-2 users at peak.