It’s covered in the Phoronix’s review
The expansion bay spec is rated for 210w at 20v (and 238.1w across all voltages), so exceeding that would require a power jack on the module. Practically I doubt that a dGPU above 200w will ever be made for the Framework 16.
An RTX 4090 mobile (which is actually a desktop 4080 except tuned for significantly less power consumption and slightly less performance) is rated at 150w and AMD’s new RX 7900m (which is a variant of the 7900 GRE except with some cores disabled and tuned for less power consumption and slightly less performance) is rated at 180w. So both of those are potential options. Their successors will likely also be potential options. But I doubt we’ll see standard desktop GPUs in a Framework laptop anytime soon.
Quite frankly, when operating beyond this power I would wonder if the current thermal system would handle it.
It would require significantly larger fans, resulting in a thicker GPU module. I suspect it may even require two sets of fans, one set to put air through the existing CPU radiator, and a second set for the GPU processor.
Can I use two charger spontaneously? Like one 180w and one 65w
No, you can’t.
No, I’m sure it can’t. Would have to be a larger module, sticking out further and possibly even adding thickness. Probably both. And even then, the fan noise is probably going to be pretty loud.
Either that or some breakthrough in cooling or reduction in waste heat, which would be amazing. That’ll happen when practical cold fusion happens. So…no time we can even guess at.
Something interesting to consider:
The interposer is almost only defined by the mainboard and the expansion module. The actual chassis itself has very little influence on the shape and capabilities of the connection, except for the two screws that hold it in place.
So what this could mean in a dreamlike utopian future, is an interposer V2 (or Interposer Pro Max Plus, to go with naming schemes in the industry) that can handle more power and data than currently possible.
That would maybe break some backwards compatibility, so you would need a new mainboard and expansion module at the same time. But maybe they can even figure that out somehow (one can dream, right?)
That being said, I believe most mobile GPUs probably never break those 210W anyway, which is probably why they chose that limit. So you would have to want to use an actual desktop card to need more power, at which point a laptop might just not be the right form factor for you. Or maybe an eGPU in an enclosure with TB/Oculink is the better way.
Edit: I know the interposer is held in by 4 screws, but there are two screws for the expansion module on either side, which limits the width of the interposer. Those are the two screws i meant.
I’m sad, no Adam Savage
Hopefully batch 5 goes out by end of 2024. The wait’s just growing depressing now…
From the newsletter:
we expect to fulfill all current batches before the end of the first half of this year.
Unless there’s some major showstopper that gets revealed in batch1/2, my guess is it’ll be no problem, probably by 1st or 2nd week in March you’ll get the “getting ready, pay up” email. A complete wild ass guess, but from what I’ve gleaned elsewhere likely.
They have said, as of a few weeks ago, that unless something showstopper comes up during initial mass production or first units to customers (aka batch 1 and/or 2), they expect to get through all of the pre-orders (at that time) by end of June or early July, if I remember right. So I think you’ll get yours pretty soon.
Update from The Verge:
ngl I dont think I’m completely sure what he means by “I was a little frustrated to see the company’s CEO suggest I only encountered one”
I think the implication by The Verge is that they had more “instability” issues in addition to the blue screen issues with the watchdog violation error. They mentioned “other” freezes, in addition to the watchdog issue. nrp mentioned that this issue was fixed and was “the” issue The Verge ran into.
I don’t know that nrp was intending to say that was definitively the “only” issue The Verge ran into. But I imagine that statement is what they were referring to.
We’ve clarified this in our Framework Laptop 16 shipping update, changing “the” to “an”.
Definitely not from anyone that still wants to work with nvidia in the future, nvidia explicitly doesn’t allow desktop gpus in laptops, even in edge cases like the egpu for the ally.
Noone said an expansion bay can’t have it’s own power input, the 210w is only what can go between laptop and expansion bay. I would be rediculous and unpractical and kinda awesome but it could be done.
I think they meant a universal enclosure with PCIe slots. And there is no way nvidia could control what consumers (or even prebuilts) put in those
But they can deny support on those in case of trouble.
that is true. I have no experience with eGPU enclosures, but they already exist and aren’t that uncommon. Does nvidia deny support to those consumers who already have an enclosure today?
I don’t really like nvidia either, but why are we trying to paint them in a particularly bad way here?
Then the whole discussion about the expansion bay shared power isn’t really relevant so I assumed they wanted to actually put the gpu in the expansion bay.
You were always kinda able to use desktop gpus in egpu enclosures as a consumer even if nvidia did try to prevent it, member good old error 43?
As a prebuilt though it’s a bit more complicated. Gigabyte used to make prebuilt egpus with desktop nvidia cards in them but asus was not allowed to use a desktop gpu in the egpu for the ally, though that may have geen because it isn’t using thunderbolt or cause nvidia got stricter with the rules.