- cross-posted to:
- framework@lemmy.ml
- cross-posted to:
- framework@lemmy.ml
I don’t get the point. Framework laptops are interesting because they are modular but for desktop PCs that’s the default. And Framework’s PCs are less modular than a standard PC because the RAM is soldered
That makes no sense - that’s more like Apple then…
I don’t know if it’s the case, but modular IO on PC maybe nice.
Soldered on ram and GPU. Strange for Framework.
Apparently AMD couldn’t make the signal integrity work out with socketed RAM. (source: LTT video with Framework CEO)
IMHO: Up until now, using soldered RAM was lazy and cheap bullshit. But I do think we are at the limit of what’s reasonable to do over socketed RAM. In high performance datacenter applications, socketed RAM is on it’s way out (see: MI300A, Grace-{Hopper,Blackwell},Xeon Max), with onboard memory gaining ground. I think we’ll see the same trend on consumer stuff as well. Requirements on memory bandwidth and latency are going up with recent trends like powerful integrated graphics and AI-slop, and socketed RAM simply won’t work.
It’s sad, but in a few generations I think only the lower end consumer CPUs will be possible to use with socketed RAM. I’m betting the high performance consumer CPUs will require not only soldered, but on-board RAM.
Finally, some Grace Hopper to make everyone happy: https://youtube.com/watch?v=gYqF6-h9Cvg
Sound like a downgrade to me I rather have capability of adding more ram than having a soldered limited one doesn’t matter if it’s high performance. Especially for consumer stuff.
Looking at my actual PCs built in the last 25 years or so, I tend to buy a lot of good spec ram up front and never touch it again. My desktop from 2011 has 16GB and the one from 2018 has 32GB. With both now running Linux, it still feels like plenty.
When I go to build my next system, if I could get a motherboard with 64 or 128GB soldered to it, AND it was like double the speed, I might go for that choice.
We just need to keep competition alive in that space to avoid the dumb price gouging you get with phones and Macs and stuff.
Sounds like a load of bullshit to feed useful idiots.
There’s even the next iteration already happening: Cerebras is maling wafer-scale chipa with integrated SRAM. If you want to have the highest memory-bandwith to your cpu core it has to lay exactly next to it ON the chip.
Ultimately RAM and processor will probably be indistinguishable with the human eye.
I definitely wouldn’t mind soldered RAM if there’s still an expansion socket. Solder in at least a reasonable minimum (16G?) and not the cheap stuff but memory that can actually use the signal integrity advantage, I may want more RAM but it’s fine if it’s a bit slower. You can leave out the DIMM slot but then have at least one PCIe x16 expansion slot. A free one, one in addition to the GPU slot. PCIe latency isn’t stellar but on the upside, expansion boards would come with their own memory controllers, and push come to shove you can configure the faster RAM as cache / the expansion RAM as swap.
Heck, throw the memory into the CPU package. It’s not like there’s ever a situation where you don’t need RAM.
All your RAM needs to be the same speed unless you want to open up a rabbit hole. All attempts at that thus far have kinda flopped. You can make very good use of such systems, but I’ve only seen it succeed with software specifically tailored for that use case (say databases or simulations).
The way I see it, RAM in the future will be on package and non-expandable. CXL might get some traction, but naah.
Couldn’t you just treat the socketed ram like another layer of memory effectively meaning that L1-3 are on the CPU “L4” would be soldered RAM and then L5 would be extra socketed RAM? Alternatively couldn’t you just treat it like really fast swap?
Wrote a longer reply to someone else, but briefly, yes, you are correct. Kinda.
Caches won’t help with bandwidth-bound compute (read: ”AI”) it the streamed dataset is significantly larger than the cache. A cache will only speed up repeated access to a limited set of data.
Could it work?
Yes, but it would require:
- A redesigned memory controller capable of tiering RAM (which would be more complex).
- OS-level support for dynamically assigning memory usage based on speed (Operating systems and applications assume all RAM operates at the same speed).
- Applications/libraries optimized to take advantage of this tiering.
Right now, the easiest solution for fast, high-bandwidth RAM is just to solder all of it.
Using it as cache would reduce total capacity as cache implies coherence, and treating it as ordinary swap would mean copying to main memory before you access it which is silly when you can access it directly. That is you’d want to write a couple of lines of kernel code to use it effectively but it’s nowhere close to rocket science. Nowhere near as complicated as making proper use of NUMA architectures.
The cache hierarchy has flopped? People aren’t using swap?
NUMA also hasn’t flopped, it’s just that most systems aren’t multi socket, or clusters. Different memory speeds connected to the same CPU is not ideal and you don’t build a system like that but among upgraded systems that’s not rare at all and software-wise worst thing that’ll happen is you get the lower memory speed. Which you’d get anyway if you only had socketed RAM.
Yeah, the cache hierarchy is behaving kinda wonky lately. Many AI workloads (and that’s what’s driving development lately) are constrained by bandwidth, and cache will only help you with a part of that. Cache will help with repeated access, not as much with streaming access to datasets much larger than the cache (i.e. many current AI models).
Intel already tried selling CPUs with both on-package HBM and slotted DDR-RAM. No one wanted it, as the performance gains of the expensive HBM evaporated completely as soon as you touched memory out-of-package. (Assuming workloads bound by memory bandwidth, which currently dominate the compute market)
To get good performance out of that, you may need to explicitly code the memory transfers to enable prefetch (preferably asynchronous) from the slower memory into the faster, á la classic GPU programming. YMMW.
I wasn’t really thinking of HPC but my next gaming rig, TBH. The OS can move often accessed pages into faster RAM just as it can move busy threads to faster cores, gaining you some fps a second or two after alt-tabbing back to the game after messing around with firefox. If it wasn’t for memory controllers generally driving channels all at the same speed that could already be a thing right now. It definitely already was a thing back in the days of swapping out to spinning platters.
Not sure about HBM in CPUs in general but with packaging advancement any in-package stuff is only going to become cheaper, HBM, pedestrian bandwidth, doesn’t matter.
The thing is, consumers didn’t push Nvidias stock sky high, AI did. Microsoft isn’t pushing anything sane to consumers, Microsoft is pushing AI. AMD, Intel, Nvidia and Qualcomm are all pushing AI to consumers. Additionally, on the graphics side of things, AMD is pushing APUs to consumers. They are all pushing things that require higher memory bandwidth.
Consumer will get ”trickle down silicon”, like it or not. Out of package memory will die. Maybe not with you next gaming rig, but maybe the one after that.
In systems where memory speed are mismatched, the system runs at the slowest module’s speed. So literally making the soldered, faster memory slower. Why even have soldered memory at that point?
Honestly I upgrade every few years and isually have to purchase a new mobo anyhow. I do think this could lead to less options for mobos though.
I get it but imagine the GPU style markup when all mobos have a set amount of RAM. You’ll have two identical boards except for $30 worth of memory with a price spread of $200+. Not fun.
I don’t think you are wrong, but I don’t think you go far enough. In a few generations, the only option for top performance will be a SoC. You’ll get to pick which SoC you want and what box you want to put it in.
the only option for top performance will be a SoC
System in a Package (SiP) at least. Might not be efficient to etch the logic and that much memory onto the same silicon die, as the latest and greatest TSMC node will likely be much more expensive per square mm than the cutting edge memory production node from Samsung or whatever foundry where the memory is being made.
But with advanced packaging going the way it’s been over the last decade or so, it’s going to be hard to compete with the latency/throughout of an in-package interposer. You can only do so much with the vias/pathways on a printed circuit board.
You are correct, I’m referring to on package. Need more coffee.
No, I don’t think you owe an apology. It’s a super common terminology almost to the point where I wouldn’t really even consider it outright wrong to describe it as a SoC. It’s just that the blurred distinction between a single chip and multiple chiplets packaged together are almost impossible for an outsider to tell without really getting into the published spec sheets for a product (and sometimes may not even be known then).
It’s just more technically precise to describe them as SiP, even if SoC functionally means something quite similar (and the language may evolve to the point where the terms are interchangeable in practice).
Ye the soldered ram is for sure making me doubt framework now.
Signal integrity is a real issue with dimm modules. It’s the same reason you don’t see modular VRAM on GPUs. If the ram needs to behave like VRAM, it needs to run at VRAM speeds.
Then don’t make it work like that. Desktop PCs are modular and Framework made a worse product in terms of modularity and repairability, the main sales of Framework. Just, like… wtf. This Framework product is cursed and shouldn’t exist.
There’s little point in framework selling a conventional desktop.
I guess they could have made another laptop size with the the dev time, but… I dunno, this seems like a niche that needs to be filled.
This is where I’m at. The Framework guy was talking about how very few companies are using this AMD deal because the R&D to add it to existing models wasn’t very viable, you really only have the Asus Z13 so I feel like being ahead of the game there will be a benefit in the long run as far as their relationship with AMD. Plus they’re also doing a 12-in laptop now as well, so it’s not like they committed all their resources to this.
Apparently AMD wasn’t able to make socketed RAM work, timings aren’t viable. So Framework has the choice of doing it this way or not doing it at all.
In that case, not at all is the right choice until AMD can figure out that frankly brain dead easy thing.
“brain dead easy thing”… All you need is to just manage signal integrity of super fast speed ram to a super hungry state of the art soc that benefits from as fast of memory as it can get. Sounds easy af. /s
They said that it was possible, but they lost over half of the speed doing it, so it was not worth it. It would severely cripple performance of the SOC.
The only real complaint here is calling this a desktop, it’s somewhere in between a NUC and a real desktop. But I guess it technically sits on a desk top, while also being an itx motherboard.
Oh yeah I’m sure you could’ve done it no problem
Not strange at all.
They’re a business that makes its money off of selling hype to morons.
Just buy a ThinkPad, if you’re thinking about buying a Framework…
“To enable the massive 256GB/s memory bandwidth that Ryzen AI Max delivers, the LPDDR5x is soldered,” writes Framework CEO Nirav Patel in a post about today’s announcements. “We spent months working with AMD to explore ways around this but ultimately determined that it wasn’t technically feasible to land modular memory at high throughput with the 256-bit memory bus. Because the memory is non-upgradeable, we’re being deliberate in making memory pricing more reasonable than you might find with other brands.”
😒🍎
Edit: to be clear, I was only trying to point out that “we’re being deliberate in making memory pricing more reasonable than you might find with other brands” is clearly targeting the Mac Mini, because Apple likes to price-gouge on RAM upgrades. (“Unamused face looking at Apple,” get it? Maybe I emoji’d wrong.) My comment is not meant to be an opinion about the soldered RAM.
To be fair it starts with 32GB of RAM, which should be enough for most people. I know it’s a bit ironic that Framework have a non-upgradeable part, but I can’t see myself buying a 128GB machine and hoping to raise it any time in the future.
If you really need an upgradeable machine you wouldn’t be buying a mini-PC anyways, seems like they’re trying to capture a different market entirely.
My biggest gripe about non replaceable components is the chance that they’ll fail. I’ve had pretty much every component die on me at some point. If it’s replaceable it’s fine because you just get a new component, but if it isn’t you now have an expensive brick.
I will admit that I haven’t had anything fail recently like in the past, I have a feeling the capacitor plague of the early 2000s influenced my opinion on replaceable parts.
I also don’t fall in the category of people that need soldered components in order to meet their demands, I’m happy with raspberry pis and used business PCs.
You can get an MS-A1 barebones from minisforum right now for like 215 - BYO cpu, ddr5, and m2. But it’s got oculink on the back (the pcie dock is 100, but not mandatory if you’re not going to use it). I think it’s supposed to be on sale for another couple days.
seems like they’re trying to capture a different market entirely.
Yes that’s the problem.
That they want to sell cheap ai research machines to use for workstation?
That’s a poor attempt to knowingly misrepresent my statement.
No, it is a question
The answer is that they’re abandoning their principles to pursue some other market segment.
Although I guess it could be said to be like Porsche and Lamborghini selling SUVs to support the development of their sports cars…
I don’t understand how that answers my question
According to the CEO in the LTT video about this thing it was a design choice made by AMD because otherwise they cannot get the ram speed they advertise.
Which is fine, but there was no obligation for Framework to use that chip either.
Suppose the counter is that the market is chock full of modular options to build a system without framework.
In the laptop space, it’s their unique hook in a market that is otherwise devoid of modularity. In the desktop space, even the mini itx space, framework doesn’t really need to be serving that modularity requirement since it is so well served already. It might make it so I’m likely to ignore it completely, but I’m not going to be super bothered when I have so many other options
deleted by creator
There’s camm2, the new standard for high speed removable memory. Asus already has released a motherboard that uses it and it matches the 8000 mts of the Framework which won’t be out until 3Q this year.
Framework chose non upgradable because it was easier/cheaper. That’s fine except Framework’s entire marketing has been built around upgradeable hardware.
Well, more specifically: why didn’t they try to go for LPCAMM?
From what I understand, they did try, but AMD couldn’t get it to work because of signal integrity issues.
Because you’d get like half the memory bandwidth to a product where performance is most likely bandwidth limited. Signal integrity is a bitch.
I thought LPCAMM was designed specifically to address the bandwidth and connectivity issues that crop up around high-bandwidth + low-voltage RAM?
Yeah hugely disappointed by this tbh. They should have made a gaming capable steam machine in cooperation with valve instead :)
This is an AI chip designed primarily for running AI workflows. The fact that it can game is secondary
Yeah exactly, its worthless… Even the big players already admit to the AI hype being over. This is the worst possible thing to launch for them, its like they have no idea who their customers are.
The AI hype being over doesn’t mean no one is working on AI anymore. LLMs and other trained models are here to stay whether you like it or not.
I mean, it’s not. You can do aí workflows with this wonderful chip.
If you wanna game, go buy nvidia
Would 256GB/s be too slow for large llms?
It runs on the gpu
Many LLM operations rely on fast memory and gpus seem to have that. Even though their memory is soldered and vbios is practically a black box that is tightly controlled. Nothing on a GPU is modular or repairable without soldering skills(and tools).
Huh?
Framework releasing a Mac Mini was certainly not on my bingo card for this year.
Ok, should I know who framework is? I’ve been a PC gamer since forever and I’ve never heard of this company.
They make repairable laptops.
repairable and upgradable*
I know it’s an absolutely banal nitpick, but I think it’s unfortunately a revelation in the current laptop market that ~90% of a laptop stays good for a really really long time, and the other 10% can be upgraded piecemeal as the need arises. Obviously this was never news to the Desktop world, but laptop manufacturers got away with claiming this was impossible for laptops in the name of efficiency and portability.
its modular like a desktop pc. you can fix it and upgrade piecemeal instead of junking it, also like a desktop. if you are a gamer you dont need to be in that common situation where the cpu still holds but the gpu is already oooolddd that usually happens on laptops.
I wasn’t prepared. I’ve been eyeing a mini for a while and this thing kills it on value compared to what I would get in a similar price point.
What alternatives were you considering, and how does the product from Framework compare?
Mac mini and studio. The overall power comparison remains to be seen but cost to spec ratio I would have had to spend over 6k and couldn’t have 16tb of memory, frameworks was around 3200.
Calling it a gaming PC feels misleading. It’s definitely geared more towards enterprise/AI workloads. If you want upgradeable just buy a regular framework. This desktop is interesting but niche and doesn’t seem like it’s for gamers.
I think it’s like Apple-Niche
It’s kinda cool but seems a bit expensive at this moment.
For the performance, it’s actually quite reasonable. 4070-like GPU performance, 128gb of memory, and basically the newest Ryzen CPU performance, plus a case, power supply, and fan, will run you about the same price as buying a 4070, case, fan, power supply, and CPU of similar performance. Except you’ll actually get a faster CPU with the Framework one, and you’ll also get more memory that’s accessible by the GPU (up to the full 128gb minus whatever the CPU is currently using)
I swear, you people must be paid to shill garbage.
Always a response for anyone who has higher standards, lol.
“It’s too expensive”
“It’s actually fairly priced for the performance it provides”
“You people must be paid to shill garbage”
???
Ah yes, shilling garbage, also known as: explaining that the price to performance ratio is just better, actually.
These little buggers are loud, right?
Hmm, probably not. I think it just has the single 120mm fan that probably doesn’t need to spin up that fast under normal load. We’ll have to wait for reviews.
I also just meant given the size constraints in tiny performance PCs. More friction in tighter spaces means the fans work harder to push air. CPU/GPU fans are positioned closer to the fan grid than on larger cases. And larger cases can even have a bit of insulation to absorb sound better. So, without having experimented with this myself, I would expect a particularly small and particularly powerful (as opposed to efficient) machine to be particularly loud under load. But yes, we’ll have to see.
The Noctua fan option should be pretty quiet.
I have a Noctua fan in my PC. Quiet AF. I don’t hear it and it sites beside me.
Question about how shared VRAM works
So I need to specify in the BIOS the split, and then it’s dedicated at runtime, or can I allocate VRAM dynamically as needed by workload?
On macos you don’t really have to think about this, so wondering how this compares.
On my 7800, it’s static. The 2GB I allocate is not usable for the CPU, and compute apps don’t like it “overflowing” past that.
This is on Linux, on a desktop, ASRock mobo. YMMV.
It’s typically dynamic
It will most likely be dynamic, with the option to statically set it.
Lmao the news about this desktop is strangling their website to the point of needing a 45 minute waiting list
They did announce three major products today.
Yeah that touchscreen tablet convertible machine is what has me psyched. I’m not the target for it, and already own a 16, but I could see that thing selling well. I honestly think they came out with the desktop because they just kinda felt they needed a desktop.
Guilty. This thing came out at the perfect time and I was considering building my own or a Mac mini but this has 95% of what I’m looking for for less than a spec compromised Mac mini. So I preordered. And I kept hitting refresh lol.
The Lemmy Lick strikes again!
Not really sure who this is for. With soldered RAM is less upgradeable than a regular PC.
AI nerds maybe? Sure got a lot of RAM in there potentially attached to a GPU.
But how capable is that really when compared to a 5090 or similar?
The 5090 is basically useless for AI dev/testing because it only has 32GB. Mind as well get an array of 3090s.
The AI Max is slower and finicky, but it will run things you’d normally need an A100 the price of a car to run.
But that aside, there are tons of workstations apps gated by nothing but VRAM capacity that this will blow open.
Useless is a strong term. I do a fair amount of research on a single 4090. Lots of problems can fit in <32 GB of VRAM. Even my 3060 is good enough to run small scale tests locally.
I’m in CV, and even with enterprise grade hardware, most folks I know are limited to 48GB (A40 and L40S, substantially cheaper and more accessible than A100/H100/H200). My advisor would always say that you should really try to set up a problem where you can iterate in a few days worth of time on a single GPU, and lots of problems are still approachable that way. Of course you’re not going to make the next SOTA VLM on a 5090, but not every problem is that big.
… but only OpenCL workloads, right?
Not exactly. OpenCL as a compute framework is kinda dead.
What types of compute can you run on an AMD GPU today?
Most CUDA or PyTorch apps can be run through ROCM. Your performance/experience may vary. ZLUDA is also being revived as an alternate route to CUDA compat, as the vast majority of development/intertia is with CUDA.
Vulkan has become a popular “community” GPU agnostic API, all but supplanting OpenCL, even though it’s not built for that at all. Hardware support is just so much better, I suppose.
There are some other efforts trying to take off, like MLIR-based frameworks (with Mojo being a popular example), Apache TVM (with MLC-LLM being a prominent user), XLA or whatever Google is calling it now, but honestly getting away from CUDA is really hard. It doesn’t help that Intel’s unification effort is kinda failing because they keep dropping the ball on the hardware side.
To anyone complaining about non-replaceable RAM: This machine is for AI, that is why.
Think of it like a GPU wirh a CPU on the side, vs the other way around.
Inference requires very fast ram transfer speed, and that is only possible (currently) on soldered buses. Even this is pretty slow at 256Gb/s, but it’s RAM size of 96GB to GPU makes it interesting for larger models.
and more at people who want the smallest, most powerful desktop they can build
Well, there’s this:
Yeah, the screw holes didn’t fit, that’s why. And the cooler didn’t fit the case, obviously. And the original cooler not the CPU’s turbo. It’s fine, it still runs most games in 3k on the iGPU.
The bowing on that board makes me think it’s not much longer for this world.
It’s like this even if i lay it on the desk. Was always like this. ¯\_(ツ)_/¯
This is not really that interesting and kinda weird given the non-upgradability, but I guess it’s good for AI workloads. It’s just not that unique compared to their laptops.
I’d love a mid-tower case with swappable front panel I/O and modular bays for optical drives; would’ve been the perfect product for Framework to make IMO.
They’d be competing with a bajillion other case makers. And I’m pretty sure there are already cases with what you ask (such as 5.25 bay mounted IO running off USB headers, at least).
Like… I don’t really see what framework can bring making a case. Maybe it could be a super SFF mobo with a GPU bay, but that’s close to what they did here.
There may be already such a case but you and me have never heard about it and it’s probably by some chinese no-name brand.
A proper metal mid-tower case with modular front panel I/O (using Framework’s system with the USB-C converters) and modular optical drive/hard drive bays would be unique.
It’s just not that unique compared to their laptops.
This’ll be a good sell for the useful idiot crowd that has been conditioned to think gaming laptops are the devil.
Really framework ? Soldered ram ? How dissapointing
The CEO of Framework said that this was because the CPU doesn’t support unsoldered RAM. He added that they asked AMD if there was any way they could help them support removable memory. Supposedly an AMD engineer was tasked with looking into it, but AMD came back and said that it wasn’t possible.
Specifically AMD said that it’s achievable but you’ll be operating at approx 50% of available bandwidth, and that’s with LPCAMM2. SO/DIMMs are right out of the running.
Mostly this is AMDs fault but if you want a GPU with 96-110 GBs of memory you don’t really have a choice.
Now, can we have a cool European company doing similar stuff? At the rate it’s going I can’t decide whether I shouldn’t buy American because I don’t want to support a fascist country or because I’m afraid the country might crumble so badly that I can’t count on getting service for my device.
I could envision MNT Research trying this in the future, but not for now.