Framework 16 was sold as a potential “gaming laptop”. But to this day, the best GPUs we can get are a 70-class nvidia gpu and a radeon 7700S… It’s somewhat, underwhelming?
Where are the high end GPU modules? Are you really just going to have 1 low-mid end gpu from each vendor per generation?
Having this GPU module slot means nothing if I can’t buy a better GPU to put in it. Where is the RX 7900M or NVIDIA 5090/4090 mobile option? Thanks.
This is a very good question. I’d gladly pay for a graphics module that packs a bit more VRAM if nothing else. 8GB is only barely enough for certain high-end games these days. That’s one of several reasons why I’ve so far declined to get the RTX 5070 module, as it has the same amount of VRAM and doesn’t seem to present much of a performance increase over the RX 7700S.
Agreed. I think of it as a future-proofing thing, but the current lack of available significant upgrades essentially means it serves no real purpose.
This is all speculation without input from an employee, but I think the reason why they haven’t been able to release as many GPU upgrades as would be ideal is because their workforce is relatively small and designing a whole GPU module + coordinating support for each new configuration is a huge amount of work. The reason why VRAM is so small is probably due to its current price.
I upgraded from 7700S to 5070. I’m obviously underwhelmed about VRAM, but the new card is way more performing than the 7700S. I’m talking about 30-40% depending on the title.
The 5070 allows me to play anything at the display native resolution at >60, and with some trickery (smooth motion) you easily pass the 100s in more demanding games, sometimes reaching the 165 mark to fully use the display.
The Framework 16 only supports 240W power via USB-C PD (which is really only 205W due to a safety throttle). Total system power with both the RX 7700S and RTX 5070 can exceed 205W easily, and so even if they came out with a higher power module, the chances of that being a good experience are slim IMO. The battery would drain exceedingly while gaming.
Well, in my books a laptop with a 70-class gpu qualifies as gaming. They explained why they couldn’t do more vram in their Q&A, I believe it’s not the price, its space limitations and configuration of VRAM chips meaning they wouldn’t be able to find a satisfactory solution to fit it in.
If it had just been a 12gb 5070 it would have been perfect in my opinion.
The point of an upgradeable GPU is that you can upgrade your 7700s to whatever GPU they will have in two years time, which will give you a huge performance uplift.
(Not to mention that if your GPU dies, it doesn’t require replacing the entire PC, and you can even still use your pc without the GPU.)
Well said
If anybody wants a beefier GPU, then buy an eGPU enclosure and go off to the races. No need to worry about power or spaces constraints then.
I know bandwidth can be a problem but that’s the only other option available at the moment.
Interesting. All the other info I’ve seen about the 5070 led me to believe it would be a dubious upgrade at best, e.g. estimates of maybe 15% better performance at most.
That being said, the other major reason why I haven’t gotten the 5070 module is because I have a deep hatred for NVIDIA as a company right now. CEO Jensen Huang’s apparent support of Trump notwithstanding, charging a minimum of $2k for a high-end graphics card (5090) is absolutely insane, and I refuse to support any company that does that sort of crap, even indirectly.
(Apologies if part of that statement gets too far into political territory.)
I have a similar sentiment towards Nvidia right now as well, but unfortunately AMD appears to be following in their tracks and trying to get a piece of the same pie. They’re not currently as bad as nvidia, but their leadership is definitely not the consumer/community driven underdog many people would like them to be.
I just want to briefly mention that some people probably aren’t aware that more powerful GPUs also require a more powerful power supply. As we already know, Framework 16, regardless of whether it’s a 7700 Super or an RTX 5070, uses almost the full power of a 240-watt power supply. That’s the current limit in terms of power consumption. As long as a more powerful power supply isn’t available, there won’t be a better GPU module either. And let’s be honest: an RTX 5080 with only a 100-watt TDP is shooting itself in the foot.
Also just in case people don’t know, this doesn’t mean we will never get a better GPU in absolute performance terms.
We will get better GPUs than the current gpus, just that it’ll be when we make a generational leap, e.g a 6070 or a next gen Radeon GPU if they decide to make new ones, fitting better performance within the same power budget 
For me, even with some AAA gaming, the 7700s is “fine” - and for some light AI use is also “fine” - its definitely low-mid range at this point in its lifecycle. But I have played Borderlands, Helldivers, etc and have no complaints.
I think I am in the same boat as others in that I would sacrifice some thickness/size of the expansion bay for a bit more RAM though. For me, I would love to get in the 12-16gb range - and would pay a bit more for a “drop in” option.
Even getting a mid-range egpu setup is going to cost about 700-1000 - so if I could get a 12-16 card that is just a drop in, in that price range, I would in a heartbeat.
I’ve been trying to figure out the power supply thing – in the configurator for the F16 with the older 7040-series CPU, there is only an option for 180W power brick. It’s the same GPU module option as the F16 with the AI series CPU, but that configurator gives the option for the 240W power supply.
Any ideas why the 240W brick is not available for the 7040-series based F16? Would I be doing myself a huge disservice going with the 7040-based F16 over the AI 300 series simply because of the power supply + GPU combo?
That really sounds like a bug.
The shop has had bugs & other oddities before.
I recently upgraded from the 7700 to the 5070 and just having the option is really a great improvement that you dont have anywhere else. Due to my work with AI development I needed a nvidia card and I was super happy they finally had that option! I would have been completely locket to that one gerenration of GPUs at launch with any other manufacturer.
I agree that the selection could be bigger, and I am confident it will be expanding. However you will need to give them some more time.
Also bare in mind that this is still a Laptop form factor, the fact that they were able to make a non-mobile GPU running is already impressive, though I would not have minded more VRAM as well …
For myself, happy with the option so far and will look out for future upgrade paths.
Not sure what you mean by this. Their RTX 5070 is the laptop/mobile version of the 5070.
Ah jea seems to me my error, hat the naming wrong. I am still used to the old m naming schemes, but its definitely a mobile version GPU.
In your defense, they dropped the m specifically to confuse people in this way.
it is very annoying when i was looking at computers before i bought one i hit the same issue!