We architected the mainboard to maximize adaptability to future generations of x86 and ARM (and we hope eventually RISC-V!) CPUs.
Maybe the more relevant quote from that article is
One of the core design principles of the Framework Laptop is performance upgradability. Not only are the memory and storage replaceable, but the entire mainboard can be removed and replaced with any of the compatible ones weâll be building in the same form factor.
The core design of the Framework Laptop is that you donât have to replace the whole laptop when the new stuff comes out so people who are asking for touch screens, or different processor architecture options, or even keyboard layouts should be able to comfortably order now if they are okay with missing that thing for a while.
I consider that Framework have addressed x86 with the Intel chips. Better IMO that they need keep up to date with only one brand of x86 and put any surplus resources towards genuinely different architectures such as Arm, OpenPOWER or RISC-V.
I get that Framework is a small company and canât just pump out new motherboard variants because some users want them, but hopefully itâs at least something being considered for the future. Also, I assume there will have to be new motherboards every year or two, because newcomers to the Framework will want equivalent performance to competing machines.
AMD have better onboard GPU options than Intel, and are also more power efficient, which should hopefully allow for a quieter machine, as well as providing more options for gaming or light CAD. However, I donât know what AMD would mean for the Thunderbolt connectivity. USB 4 isnât going to be back compatible with Thunderbolt specific hardware, is it?
Again, check the wording. âfuture x86 platformsâ. That could easily just mean intel 12th or 13th gen. Theyâve been very coy about any AMD hardware on the platform and I find that disappointing. Iâd love to hear them respond but theyâve also been super quiet in AMD threads as well
I would much prefer 1920Ă1280 to the current resolution. Maybe I just need to see it in action, but 2256Ă1504 feels like an odd-ball resolution that would need decent 1.5x scaling (which⊠doesnât exist on Linux) to not look like trash. I know âgamersâ claim âyou donât need more than maybe 1440pâ, but honestly, while that may be true for gaming and content consumption, since COVID Iâve been using a 4k monitor and 1080p looks like trash now. (And once upon a time, I was used to 640Ă480 .)
My current laptop runs 3200Ă1600, which is usable if not ideal, and very similar to 3000Ă2000 panels available in the 3:2 13" form factor. Seriously, though, for productivity, higher resolution looks great and IMHO is actually more beneficial at smaller physical sizes as it improves text clarity, which is more important (not less) when what youâre trying to read is already tiny.
I would absolutely prefer AMD (better performance, better iGPU, better price, and even better power consumption from what Iâve heard), but itâs not my personal deal-breaker. The screen and keyboard are, and Iâm most suspicious of the keyboard in particular ever being upgradable given the chassis integration. (Agreed on â„32 GB RAM as a requirement, though!)
âŠexcept that the entire innards (minus keyboard) have to be moved to a new chassis. And I am not holding my breath that weâll ever get said chassis. Please, prove me wrong!
@Zax I think the reason theyâve been so broad is because they donât wanna announce anything too soon while something is too early in negotiations/development. They could be also just focusing more on non-X86 platforms like ARM or RISC-V because most of the big players are banking on those being the future, especially ARM.
So like, thatâs a fair point but I donât see ARM being dominant on the windows side for a while. Too much software needs x86 and windows doesnât have a good ârosettaâ style tool to run it. Also, qualcomm is slow as all hell compared to apple M1. Thatâs part of why appleâs approach works, they have a TON of power to work with. Non-Apple arm chips are pretty weak, since theyâre focused on area efficiency rather than raw power.
While the M1 is powerful, itâs worth noting that a great deal of itâs aptitude at JIT-translated x86 emulation is not simply due to power, but has a great deal to do with having special extensions to the instruction set and memory ordering behavior for that purpose specifically. Thatâs perhaps even more important than âhow powerful it isâ with respect to being able to take that approach.
Apple is also not the only game in town for large ARM chips with more focus on âraw powerâ. Amazonâs Graviton2 processor they made for AWS is supposedly about 2x the transistor count as M1 for instance. HiSiliconâs Kirin 9000 while a little slower is not too far off, roughly able 10% slower, and a comprable transistor count. There are others working on large ARM processors as well.
EDIT: Ah, 10% figure was wrong by quite a bit, was going off bad information from an overly hasty search. Still, Apple is not the only one working on bigger ARM processors, and I fully expect there to be more over time. Apple is simply a first-mover for this class of ARM processor because they have way more money than anyone else to throw at it.
@Zax While that has been the case with ARM so far, I think @AlexS pretty much nailed it. There are indications they are shifting from efficiency and weâre gonna start seeing some consumer-grade, high-powered ARM processors very soon. I think when Apple dropped the M1 and when NVIDIA had such keen interest to acquire ARM, I think it was a big indication of intent.
Iâm going to be blunt here. The graviton isnât a good example, itâs a server cpu based on an old arch designed for many cores and low power consumption.
Everyone uses ARM reference designs except apple. The result? The M1 manhandles the cortex x-1 cpus in the snapdragon 888+ and whatnot. Itâs a joke. You canât compare transistor count from a 32 core server cpu to a laptop chip, thatâs not a real metric. Iâm also not seeing the Kirin 9000 being 10% slower than an M1, itâs probably comparable to an old a13 cpu.
Until ARM decides to make some real changes to make solid laptop capable CPUs and until Windows gets their act together to make a real windows RT platform, weâre not going to have an ARM future.
AMD actually has fallen behind Intel in IGPU performance with Vega now underperforming Intelâs Iris XE graphics across the board, and there is no current Ryzen mobile hardware that supports Thunderbolt. Best it can do is 20 Gbps which isnât going to cut it for Thunderbolt 4 devices.
I donât think anyone would âpull the triggerâ on an AMD option because I donât think Framework would be able to get enough stock of Ryzen mobile processors to meet their current demand. AMD only accounts for about 18%-20% of the mobile market and their yields of mobile processors is no where near the level of Intel. Regardless of how they perform Intel beats them in just sheer volume. Current model Ryzen 5700U laptops are just now starting to come back in stock, and 5800U laptops are almost impossible to find in any significant quantity. 4000 series Ryzen mobile might have viable supply, but at this point those are already a generation old. Give AMD time to get their Mobile presence up to where their Desktop presence is and then Framework will have a viable path to integrating it into their inventory. Also, consider Framework has to get their market presence up to a point where AMD will consider allocating their inventory away from the bigger market players like HP, Acer. Lenovo, etc.
My laptop is getting old, but it works. I got excited by frame.work, but Iâm done with Intel and their lack of respect for security. When (or if) you release an AMD laptop, I would be very interested in replacing my laptop.
In addition to this, we also have to remember that laptop design is far more complex than desktop design. You donât just buy a chip and plug it into a standard motherboard - you need to design a whole new motherboard and cooling solution from scratch (made even more difficult by Frameworkâs modular design). As I understand it, Intelâs service team is outstanding, and work extensively with OEMs to design their machines.
While Iâd have loved a 5600U for my laptop, itâs not like there are a ton of those out there from competitors. Iâve looked. And waited. Even in late 2021, they remain few and far between.
Rumours suggest Intel is motoring ahead with Risc-V. Whilst itâs anyoneâs guess when these find their way into laptops, I think it might occur sooner than people appreciate. Clearly Intel wouldnât want to destroy sales of their current range by marketing a successor which could be a couple of years out.
At this point, perhaps loyalty to the vendor will have its rewards in the form of access to these chips?
Itâs also important to note Frameworkâs modularity is dependent on USB4, which to my knowledge isnât supported by AMDâs mobile processors.
Yeah! But intel release the USB4 standard, so AMD can implement it as well
I have a batch 2 i7, but Iâll be first in line for an AMD mainboard replacement if such a thing becomes available. I said in a different thread: Intel sucks, but Framework is awesome. So Iâll support them in the hope that theyâll offer alternatives soon.
It should be supported in Zen 4 next year, and 12th gen Intel (I forget the codename) looks like itâs going to be really good.
Iâm happy with Tiger Lakeâs specs, but irritated that I canât undervolt. Thatâs not a dealbreaker for me, but it does mean Iâm thinking about an upgrade next year rather than waiting two generations like Iâd planned (Iâd hoped to skip the first DDR5 release). Assuming I can just swap out the mobo and memory, thatâs another point in Frameworkâs favor.