I got a framework laptop not too long ago and I’d would love you have a better gpu in it, so I was wondering what it would cost me in order to do that.
Hi and welcome to the forum.
There is no upgrade from the 13.5" to 16"
You will have to buy a new 16"
There should be a way to buy a 16 inch chassis and just migrate everything that is interchangeable over. Perhaps this thread could help encourage that idea.
13 inch hardware cannot be transferred over, unfortunately. At most, storage, wireless, and MAYBE memory depending on the config. There exists no upgrade path, sadly.
When the Framework Laptop 16 becomes available, for the first time a different chassis will be available. Is Framework considering an upgrade kit available for moving Framework Laptop 13 components to a 16?
As far as I know, this information isn’t available yet.
My guess is the motherboard model will be a different size, so I would expect the chassis, monitor, and mobo are incompatible along with the power adapter which we already know is different. What could be compatible would likely include RAM (depending upon RAM DDR4/5 compatibility), M.2 hard drive, and the expansion cards. The keyboard may be compatible between the two, but it’s hard to know thus far.
At this point, I would say it seems reasonable you would need to buy a whole new Framework 16 DIY or Pre Built laptop kit when it becomes available. But who knows, it may be more compatible than I’m estimating here!
As coxdm said, components like RAM probably will be compatible, but interfaces like the expansion bay system are close to impossible to adapt, so it’s going to be a close to certain no.
There’s no need, you likely can buy the Framework 16 without the SSD and RAM and expansion cards, and those are the only parts that are potentially compatible. Except maybe also the Wifi card, but that only costs $18 or so and that is probably always preinstalled.
So in theory, you could build a mod to allow this. In practice, you would be unhappy with what you received. You’d be much better off selling your 13.5" and using the money towards a 16".
Let’s start with the biggest hurdle: the GPU. The FW16’s motherboard needed to be either largely or completely redesigned from the FW13 in order to accommodate for the swappable GPU design. After designing the Expansion Bay electrical footprint, they needed to make room for it on the motherboard. No pinout, no connection. No connection, no GPU. The FW13’s motherboard does not have these pins to make that connection. Even if you managed to fit a FW13 mainboard into a FW16 chassis, you would need to figure out how to connect the GPU.
You could figure out some kind of adapter. Framework showed off a proof-of-concept eGPU dock for the FW16 GPU modules. If you took the guts of that out of the enclosure and fit it into the chassis, you could be able to plug in the GPU where it’s meant to go, and connect it to the FW13 mainboard via Thunderbolt USB-C. Whether you’d get the same performance or not is questionable, though.
Framework 16’s PCIe Gen4 x8 interface can transfer up to ~16GB/s, equivalent to 128Gbit/s. USB4 and Thunderbolt 4 can transfer up to 40Gbit/s. That’s incredibly fast, but it’s about 31.25% of the speed you’d get from using the GPU module directly in a FW16 mainboard. This is still being optimistic, though. You’ll probably need to carve out some bandwidth for the DisplayPort signal, in order to actually get the image data from the GPU to your mainboard and then to your monitor.
You will likely also be limited by the power available. The FW13 charger can supply up to 60W, but the mainboard can process up to 100W. Similarly, the FW16 charger can supply up to 180W, but the mainboard can process up to 240W. If your gaming isn’t throttled by the connection bandwidth, it might end up throttled by the amount of power available.
Considering this is one of your reasons for wanting to do this transplant, this part should dissuade you. It would be a huge amount of work for an inferior experience.
Second issue is the chassis. The mainboard and chassis have holes and standoffs lined up to make it sit in place perfectly. We don’t have images of the FW16 internals yet, so it’s possible these will line up nicely between the FW13 and FW16, but I’d say it’s unlikely.
There’s also matters like thickness and cooling to consider. We don’t know the thickness of the internals yet, so we don’t know how well it would fit. This compounds issues like the eGPU adapter, which would likely be too thick to fit without significant modification.
We know a little bit about the cooling design, so you’d have more of a foothold here. Cooling is important. If you don’t put the time and effort into it, you will run into problems. If the computer runs too hot, it will automatically throttle performance in order to avoid permanent damage. And if it runs too hot for too long, the lifespan of the internals will be shortened. Figuring out the airflow would be tedious, but not impossible.
Third issue is the expansion cards, but this may actually be the easiest to address.
Firstly, the FW16 is wider than the FW13, so the IO ports wouldn’t immediately line up on both sides. This might actually be to our advantage, though. This allows us room to route a USB cable to the eGPU electronics we mentioned earlier. This also allows us room to add a USB hub (trust me, we’ll need it). The FW16 offers 6 IO ports instead of 4. Let’s say the front-left two get direct mainboard access, and we connect the other 4 via USB hub. But we’re still not done with this USB hub! You know why? Input modules!
How exactly are the FW16’s input modules connected? We don’t know yet. Do they plug in directly into the mainboard, or is there a controller board that all of them plug into that then feeds into the mainboard? We don’t know yet, but my guess is the second. And you’d better hope this is the case, because it makes our job much simpler than the first possibility. If it’s all running towards one connection, then we can plug that into our USB hub and be done with it. Well, almost done. We need to get sleep state data in order to fully comply with the protocol. I’m not sure how to do that, but you could probably figure something out given enough time.
Next, you need to connect to the display, and maybe the battery. Both of these will use different connections between FW13 and FW16. If you’re bringing over the FW13 battery, you’re good to go (after figuring out how to securely attach it). If you’re using the bigger FW16 battery, you’d better find an adapter. While you’re at it, find an adapter for the display, too.
Lastly, space and money. How much room do you have inside this chassis? Is it enough room to fit all these modifications? You can reduce the footprint of these mods by designing a printed circuit board (PCB) with all the necessary connections and circuits for all these mods. It would cost more money, but you’d end up with a better installation and maintenance experience. The downside here is just how much it takes to design a PCB for complicated processes like DisplayPort, USB, and PCIe. It can theoretically be done. But you would need to learn a lot about electrical engineering in order to do it.
Let’s say we want Framework to design that for us. Let’s say we can convince them. This kit is now available in the Marketplace. You can purchase the chassis, display, battery(?), speakers(?), and adapter board as a kit, and transplant your mainboard yourself. You have now paid hundreds of dollars for an inferior experience, and you have a FW13 shell that is no longer useful.
I’m not so sure that fits the Framework mission.
Without a mainboard to run it, that FW13 shell is e-waste. If you happened to upgrade your FW13 mainboard in the past, you can put the old one back in and have a functioning system. But I’d wager most people don’t upgrade every generation or two, and won’t have one lying around.
But hey, maybe you’ll upgrade to a FW16 mainboard in a few years. Then you’ll be able to put the FW13 mainboard back into the FW13 shell and have a working computer to sell/use again.
Fitting a FW13 mainboard into a FW16 chassis is an interesting idea. It would be a really fun project to work on, just for fun. But it would be an inferior experience, and doesn’t make a lot of sense as a business/consumer decision.