Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Framework Desktop is a beast

world.hey.com

Post reply on HN
Print view View original

211–220 of 464 posts

Re: The Framework Desktop is a beast

#211

Post by jchw »

esafak wrote: post #206
rs186 wrote: post #136

Earlier quoted context omitted.

If you can't find an sufficiently similar alternative that is priced at a much better price, it is not overpriced.

For the purposes of running LLM models, a Mac Mini. The PC is cheaper, but it doesn't have MacOS, Apple's service or resale value.

Actually the pricing is pretty similar.

Framework Desktop price with default selections, 32GB of RAM, 500 GB storage: $1,242.00 USD

Mac Mini with 32GB of RAM, 512 GB storage: $1,199.00

Post changed a bit since I started replying, so:

> For the purposes of running LLM models, a Mac Mini

The M4 Max is the one that actually gives you a shit load of memory bandwidth. If you just get a normal M4 it's not going to be especially good at that.

> it doesn't have MacOS

The Mac can't run Windows, which is used by ~75% of all desktop computer users and the main operating system that video games target. I'd say that would be the bigger problem for many.

> Apple's service

What advantage does that get you over Framework's service?

> resale value

Framework resale value has proven to be excellent by the way. Go to eBay, search "Framework Laptop", go to "Sold Items". Many SKUs seem to be retaining most of their value.

(Nevermind the ease of repair for the Framework, or the superior expandability. If you want to expand the disk space on an M4 you need to get sketchy parts, possibly solder things, and rescue your Mac with another Mac. For framework devices you plug in another M.2 card.)

Re: The Framework Desktop is a beast

#212

Post by yencabulator »

zyx321 wrote: post #197

There's been some theories floating around that the 128gb version could be the best value for on-premise LLM inference. The RAM is split between CPU and GPU at a user-configurable ratio. So this might be the holy grail of "good enough GPU" and "over 100GB of VRAM" if the rest of the system can keep up.

> The RAM is split between CPU and GPU at a user-configurable ratio.

I believe the fixed split thing is a historical remnant. These days, the OS can allocate memory for the GPU to use on the fly.

Re: The Framework Desktop is a beast

#213

Post by vaylian »

paolomainardi wrote: post #174

Is there a real alternative, where is not not needed Windows in any way, to the Remote Device Management baked into firmware as Apple does with its hardware ? This is biggest missing to bring Linux to enterprises.

> the Remote Device Management baked into firmware as Apple does with its hardware?

What do you mean? Linux had SSH (and before that rlogin) for a very long time already.

Re: The Framework Desktop is a beast

#214

Post by kcb »

DiabloD3 wrote: post #165
dgan wrote: post #107

Earlier quoted context omitted.

sooo what's the successor of cuda?

CUDA largely was Nvidias attempt at swaying Khronos and Microsoft's DirectX team. In the end, Khronos went with something based on a blend of AMD's and Nvidia's ideas, and that became Vulkan, and Microsoft just duplicated the effort in a Direct3D-flavored way. So, just use Vulkan and stop fucking around with the Nvidia moat.

A great thing about CUDA is that it doesn't have to deal with any of the graphics and rendering stuff or shader languages. Vulkan compute is way less dev friendly than CUDA. Not to mention the real benefit of CUDA which is that it's also a massive ecosystem of libraries and tools.

Re: The Framework Desktop is a beast

#215

Post by neuroelectron »

You know it would be a great use for this chip besides AI generative slop? A desktop server with AI-enhanced firewall and a silent home server. This could be effective for zero-days and other weird traffic patterns and maybe even add enterprise-ish email server with spam detection.

There has been several projects started that are experimenting with this including Suricata and pfSense. I wonder how well this chip could handle packet inspection.

Re: The Framework Desktop is a beast

#216

Post by yencabulator »

cuu508 wrote: post #177

What are more budget friendly options for similar workloads (running web app test suite in parallel)? My test suite currently runs in ~6 seconds on 9700K. Would be nice to speed it up, but maybe not for $2000 :-) Last I checked 13700K or 13900K looked like the price/performance sweet spot, but perhaps there are better options?

I bought https://store.minisforum.com/products/minisforum-um890pro to compile Rust faster than my laptop, with 96 GB RAM and 2x4 TB NVMe as a ZFS mirror. Back before Framework Desktop existed.

It has the 8945HS CPU, the article benchmarks against 8745H which is a little bit slower. It's a very worthy price point to consider, tiny and very quiet.

qwen3:30b-a3b runs locally at 23.25 tokens/sec. I know 395+ chip would about approximately double that, but I'm not quite willing to put $2000 into that upgrade.

Re: The Framework Desktop is a beast

#218

Post by worthless-trash »

fmajid wrote: post #82

I cancelled my Framework Desktop order and ordered a HP Z2 Mini G1a instead, the goal being to replace my Mac Studio as I've had it with Apple's arrogance and lousy software quality. The HP is much smaller, has ECC RAM and 10G Ethernet. Significantly more expensive, however.

Do you linux on this ? if so.. does everything work ?

Re: The Framework Desktop is a beast

#219

Post by reactordev »

>The AMD 395+ uses unified memory, like Apple, so nearly all of it is addressable to be used by the GPU.

This is why they went with the “laptop” cpu. While it’s slightly slower than dedicated memory, it allows you to run the big models, at decent token speeds.

Re: The Framework Desktop is a beast

#220

Post by justincormack »

Marsymars wrote: post #202
sliken wrote: post #124

Earlier quoted context omitted.

Keep scoffing, it's not real end to end ECC, just "link" ecc, which is just part of the chip -> CPU pipeline. So it's not full ECC like servers have with dimms with a multiple of 9 chips with ECC protecting everything from the dimms to the CPU. Keep in mind the ram is inside the strix halo package, not something HP has control over.

Wait is there any actual difference in the RAM between the HP and the Framework Desktop?

No I dont think so, all ddr5 has some "ECC" but not full ECC, ie you cant see corrections. And all the AI maxes have the same LPDDR5
Post reply on HN