I saw this video and it said that FW16 was a close collaboration with AMD: Introducing the Framework 16 AMD Advantage laptop - YouTube
Does it mean that FW has to not sell competitor GPUs even if the expansion bay can allow a different GPU? Or not sell them for some time?
Anyone who knows the answer to this question, will not be replying here. For now the FW16 is AMD based only. My suspicion is if there’s demand, this will also eventually include NVidia.
Looking briefly at AMD Advantage partners, being in that program does not prevent brands from offering Intel/NVidia options, even for identically branded devices:
https://www.amd.com/en/where-to-buy/laptops/advantage.html
Or more specifically:
https://www.lenovo.com/us/en/p/laptops/legion-laptops/legion-7-series/legion-7-gen-7-(16-inch-amd)/len101g0017
https://www.lenovo.com/us/en/p/laptops/legion-laptops/legion-7-series/legion-slim-7i-gen-7-(16-inch-intel)/len101g0020
My gut reaction is that such an agreement would be antithetical to Framework’s mission. On the other hand, AMD has apparently been very eager to work with Framework on this, while NVidia is popularly a bully to partnering businesses and seeing an NVidia branded module may take a third party to collaborate and create an expansion bay compatible GPU.
All conjecture, of course.
Nvidia is not known for their cooperative nature. See Apple vs Nvidia and the GeForce Partner Program. I think your assumption that it may take a third party, such as ASUS, Gigabyte, PNY, or some other Nvidia OEM, to bring Nvidia GPUs to the Framework 16, is a good assumption.
That being said, with the FW16 GPU module being an open design, we could see other manufacturers adopt the PCB layout and launch GPUs on multiple laptops simultaneously, just tailor the cooling solution for Alienware, HP Omen, Framework, etc,. should enough of them support the format.
I hope that there will be an Nvidia option soon, I don’t game much on my laptop so that doesn’t really matter to me but many pieces of software that I use have pretty bad AMD GPU support. Blender only just got support for AMD ray tracing through HIPRT and it’s not even close to Optix on Nvidia in terms of stability or performance.
The concept needs to be highly successful first.
My guess is that there is a lot of AMD specific technology behind the form factor functioning. Does NVIDIA or Intel have similar technologies? I think this is the first question that one should be asking.
To be honest, I’m not sure. I have a feeling that an Intel board will be made for the FW16. The $64,000 question is going to revolve around compatibility. Unless Framework manages to pull out some magic, I have a feeling that AMD CPU’s will be paired with AMD GPU’s, Intel CPU’s will be paired with Intel GPU’s, and those looking for Nvidia graphics will be down the creek without a paddle. I really hope that I’m wrong about that but others have mentioned Nvidia’s lack of willingness to play nice with others. That said, if Framework can pull off some standardization between manufacturers, we might see an Intel GPU with an AMD CPU or vice versa, through a GPU expansion. If we’re lucky, Nvidia may decide they need to get in on this and we see Nvidia graphics with either CPU. I’m patient enough to wait and see how this turns out.
I don’t see the problem here. There have been laptops with a mix of intel/amd cpus and nvidia/amd gpus for ages now. I believe it’s more or less like on any desktop PC: you install the corresponding driver for your graphics card and you’re good to go.
Laptop manufacturers also need to worry about the hardware in terms of motherboard layout and so on, but again, they have been doing that forever and i don’t think there is anything preventing framework or anyone else to slap any other gpu in the expansion bay, as long as they find a partner to build it and that probably mostly comes down to financial viability of that specific gpu model in that formfactor
The bus to the expansion bay includes PCIe 4 in the pinout. Given that PCIe 4 is the same standard used on current desktops, and that NVidia and AMD GPUs are interchangeable between platforms, why do you feel this is a limitation for FW? My suspicion is the major limitation is FW deciding to do a tie in with AMD for launch. The question is how long they gave AMD exclusivity.
Sorry, NVIDIA would rip off Framework to hell and back. Can imagine a RTX 4060 mobile module being $699, no thanks.
I think it’s interesting that framework specifically mentioned that Cooler Master collaborated on the design of the cooling system in the expansion bay modules. Cooler Master also produced a 3rd party case for the framework 13 mainboard. I could imagine a scenario where other partners like asus, or maybe a brand that doesn’t also make laptops like zotac could design a module for the system with a variety of gpu configurations since the specs are openly available, and based on open standards like PCIe.
I know it might sound crazy, but I really hope we see a Quadro for the 16.
Right now I have a couple of clients who use Quadro equipped laptops, but like most laptops serviceability is awful, and getting a committment in terms of longevity of servicing is very difficult. The idea of a laptop which can have a Quadro GPU dropped in would be brilliant, and it is a part of the market where the price would be less of an issue.
I know we are probably looking down the road quite a bit here, but I truly hope the Framework expansion port will become a standard and we see 3rd party manufacturers releasing GPU’s such as these.
Maybe pre-builds that are preconfigured will have to adhere to whatever CPU/GPU pair the chip providers want, but DIY has free reign?
I hope eventually all parts can be picked and chosen as long as they are compatible with each other on both DIY and pre-builds.
For game development, AI and simulation - nVidia GPUs are key. It’s unfortunate but the reality is that RTX platform is miles ahead in support and software than AMD.
I run a 20+ person game dev shop that likes the flexibility on working on high performance laptops - self-upgradable laptops would’ve been perfect. My team was excited about the Framework. But without nVidia GPUs, Framework is a hard pass unfortunately.
This thread seems to be veering off into the direction of another “what GPU do you want for FW16” discussion, like this one: [Poll] CPU and GPU combinations
Nothing wrong with that, but just to circle back to the original question:
Framework have put a lot of effort into creating or preserving user choice with every component they can. Their response to user feedback has been outstanding, in my opinion. I don’t know anything more than anyone else in this thread, but I think not eventually providing some kind of choice for this component would be somewhat contrary to the spirit of the project, and a bit of a departure from their excellent track record for responding to community feedback.
I think we just need to give them more time to work out the details, and more choices will emerge. After all, the laptop has not even been released yet!
While I agree that Nvidia tends to lead the pack, you’re talking about a company with a market cap of 1.13T USD vs AMD with a market cap of 171.3B USD; and let’s not even get into the actual corporate financials. Nvidia’s entire GPU division is larger than AMD in it’s entirety, it’s incredible that AMD is even as close as they are in the grand scheme of things.
And what do we get for it? Ever rising GPU prices, further market lock-in, more winning for Nvidia resulting in more winning, which results in more winning. You can see where I’m going. If you make stuff tailored for Nvidia, you lock everyone else out. I don’t doubt that there’s a reason why Unreal and modern Unity run like complete garbage on modern systems, and I have a sneaking suspicion that Nvidia is involved, given their shenanigans in the past.
Tailoring your stuff for AMD/Intel though? Works great across all sorts of systems. Sure, AMD/Intel and their push for interoperability and open-source is a marketing play, but at the end of the day, that’s what lets us, the consumer, win. Intel couldn’t have even the half-decent GPUs they have today without AMD and Valve so they could sidestep the problem Arc graphics have with older APIs.
As long as AMD and Intel have “good enough” graphics, I’m going to support them. I’ve been rolling AMD graphics since 2012, I’ve seen their bad products and skipped them, but they’re on the upswing, it just takes time before they can properly compete against an industry giant that, unlike Intel, isn’t complacent with their position.
Actually it’s frightening and concerning 
Perhaps getting FW16 laptops with TB or Oculink eGPUs? And maybe in the future, nVidia expansion cards will come out. Only if you’re really set on getting a FW16 this generation.
Just like FW13, FW had to start somewhere. They started with Intel. People asked for AMD (hard pass if no AMD! they said). So finally we have AMD. Took them 3 generations, but we’re here.
FW16 is probably going to be in the same boat. I’m actually surprised this time they started with AMD. I guess with FW13 + AMD, the door was open for a FW16 + AMD start.
I thought with a FW13 + Intel start that the FW16 would already have a head-start with Intel as well. I’m not sure why an Intel option wasn’t available as well. Maybe cost? Decide to dip their toes in the 16 inch model with one partner first? With the clamor over AMD, FW may have decided this would be a better first choice than Intel?
Though just like with the numpad situation, the Intel/AMD/Nvidia they may be learning the split is more even.
Even so, this is how things are now, but looking at how the FW13 story played out, I feel the FW16 will also have multiple choices eventually.
It’s already amazing that a small company like FW can offer Intel and AMD on a product line already. A new product line I wouldn’t have expected multiple choices. However, with their track record, I have no doubts we’ll have the choices we wanted in time, just like with the 13 inch model.
Perhaps getting FW16 laptops with TB or Oculink eGPUs?
It’s a neat idea! I don’t know how often the GPu itself would need to be portable, and even though this is kind of “lunchbox” form factor is kind of neat:
https://www.amazon.com/AKiTiO-Node-Titan-Thunderbolt-Enclosure/dp/B085TH77RT/ref=sr_1_3?keywords=egpu+thunderbolt+3&qid=1694977783&sr=8-3
However I’d suspect the bus bandwidth would be severely limiting when the data is being transmitted over thunderbolt, no? I’m no hardware expert so I would have to ask how to the bandwidth speeds would compare. I doubt it would ever get to be as fast as plugged-in into the mobo.