Fundamentally, an ARM laptop should run cooler than its Intel counterparts. This could mean that ARM laptops need to be throttled less, allowing them to run faster and do more work.
There’s been more than one thread talking about this. The modular design definitely increases the odds that an ARM swap is possible, but first there needs to be a good ARM chip that isn’t an M1. The only real game in town right now is the Snapdragon SQ2 which is not that impressive.
Really? I don’t know enough about what is available (right now) but I do feel that aarch64 seems to be the (or at least a) route to explore with the frame.work philosophy! There are some obvious advantages to this architecture
There are some disadvantages to the ARM architecture as well. For example right now each SoC has to be targeted specifically by OS developers for support. This is a huge overhead and ARM boards can “die” essentially from lack of software support. For a company like Framework that is targeting sustainability the move to ARM is going to be more complicated than it was for Apple which is both making the hardware and supporting the software.
… sure. That is true about ANY architecture. A simple solution could be to to collaborate with the Raspberry Foundation! A modified board as is in the 400 with better cooling and more RAM, clocked to, say 2.2GHz?
It’s specifically not true about the x86 architecture. It is much easier for OS developers to add support for new x86 motherboards vs new ARM SoCs.
I like your suggestion about partnering with a proven manufacturer in this space like Raspberry. That seems to me like it would be the easiest way to support ARM. You wouldn’t even need to partner all that closely someone could make a compute module IO board that fits the Framework form factor. However for that better software support you have to sacrifice performance because the Raspberry Pi Computer Module 4 SoC still doesn’t even compete with the low end x86 processors.
If your goal is to create a sustainable laptop without sacrificing performance this year then ARM just isn’t there yet. I personally hope that it gets there and Framework has mentioned that it is something they are looking at for the future. But I’m glad that they started where they did since if this laptop takes off as a platform then when it does make sense to transition I won’t have to buy an entirely new laptop.
… yes, that’s a tall order!
But the price starting at $999 could be significantly lowered … an entry level laptop in South Africa at the moment is less that half that price! If an ARM based framework laptop could compete at that level (with sacrificing a bit of performance) the buyer could be tempted to buy the Intel-mainboard later the AMD-mainboard (for gaming maybe?) and use them all in the same framework enclosure when needed!
Also, if the price of Win10(or 11) bloatware could be removed … frame.work, maybe partnering with a linux distro of their choice - imagine the possibilities !!
Well there’s already great news on this front. The Framework DIY edition is already $139-$199 less expensive if you don’t get the Windows 10 License and want to run Linux instead. No distros preinstalled yet but they have announced that they are sending laptops to distro maintainers to improve support and that may lead to partnering officially with distros in the future. I personally went for the DIY edition for exactly this reason since I don’t intend on running Windows 10 or 11 on it ever. Windows on ARM support is a whole separate set of issues. When ARM is eventually supported officially by Framework I imagine it will be Linux first.
This part is really the sticking point in our conversation (which I’ve quite enjoyed by the way). I cannot speak for Framework since I am not employed by them but it seems to me they are unwilling to sacrifice performance, it seems as important to the overall design of the laptop as the repair-ability and the Expansion Card system. Which makes a lot of sense, if people want a budget ARM laptop with decent repair-ability and reasonable performance there are already great options out there in the Pinebook Pro. Pine64 is great at providing parts for repairs and even upgrades for some of their products like the PinePhone. It’s not as modular and repairable as the Framework but it comes close enough to illustrate a point. Framework could probably release an ARM board that is equal to the performance of the Pinebook Pro but it would be a very different product to the high performance Framework Laptop for the end user. I think Framework is interested in offering users choice but not if it means altering the design of the product.
If I allow myself to dream of a bright future where lots of people adopt the Framework then this is what I envision as the future of ARM on the platform: First community members release an ARM PCB probably a Raspberry Pi Compute Module 4 based board that Framework owners who are interested can buy on the Marketplace and run in their Framework laptops with no official support from the Framework team. Then hopefully just a few years down the line the ARM ecosystem has changed and Framework can design a mainboard that runs on ARM without sacrificing the overall design of the laptop which they will officially support.
…
Excellent idea! I’ve also enjoyed your knowledgeable insights …
@Shaun_Kingston and @Rhodesian The Raspberry Pi CM4 is something I had thought of as well, and mentioned here: Raspberry Pi CM4?
@nrp replied that it’s technically possible, and that they’d be open to working with someone on it…
@Dan_Brunsink It seems to me to be the easiest option. The Compute Module 4 has a small vertical footprint now that it isn’t in the DIMM form factor like the old compute modules. So I imagine it would just take an enterprising group to make an IO board that fits the framework form factor. I do foresee some difficulties however. First I’m honestly not sure if the CM4 is capable of supporting the expansion card system and second the thermals might also be a problem as the raspberry pi 4 SoC runs surprisingly hot. But those things might not be deal breakers for people who just want to run ARM in their sustainable laptop case. Maybe a board based on the CM4 would have its own custom expansion cards also created by the community that wouldn’t be based on USB C but some other system to expand the IO of the CM4 based board. Or at least a way to make the IO ports flush with the outside of the laptop.
I’ve seen a lot of posts talking about getting AMD CPUs on the Framework laptop, but what about chips based on ARM or RISC-V? Either of those instruction sets would be able to increase the battery life significantly, let the unit run cooler, and potentially free up space from not needing a fan to be able to get a secondary NVMe slot or something else. This of course comes at the cost of software compatibility, but there are a few reasons this would not be a huge problem; namely, someone who needs x86 software would choose an x86 motherboard, 32-bit programs can be run (at least on Linux) fairly well with Box86, 64-bit program compatibility is in development with Box64, and by moving towards ARM the market can be pushed in that direction as well, leading to more natively compatible apps. I think that with the existence of Box86 and Box64, it would be a mistake to not make an affordable laptop which packs some power and can last for multiple work days.
I’d be very interested in RISC-V. Maybe we should buckle down and make one
I imagine that assuming the appropriate drivers exist for the ARM variants (trivially true for any hardware they’re currently using if Linux is the OS, as those drivers would be open source and able to be built for ARM manually; Windows is less clear), it wouldn’t be incredibly difficult. I’d definitely love an ARM variant though.
From the ‘The Upgradable Mainboard’ blog post:
“We architected the mainboard to maximize adaptability to future generations of x86 and ARM (and we hope eventually RISC-V!) CPUs”
So yes, I hope for a bright, RISC filled future! Apple already proved the power of RISC, and I hope both ARM and RISC-V continue to have more serious adoption rates. Hopefully Alder Lake and future generations of CISC processors in general get to be a lot more efficient as well!
A repairable laptop
That you can use :
Intel
Arm
Amd
Can you imagine… I hope they can accomplish this
Oh my gosh, yes, I’d take both ARM and RISC-V (well, the latter just for tinkering, since RISC-V isn’t ready for it’s prime time yet).
More realistically an AMD x86_64 option would be nice.
The big thing for me is feature parity. Performance isn’t 1:1 comparable so I care less about that, but things like Thunderbolt (USB4) are a MUST for compatibility with add-ons and docks. Currently there’s nobody selling an ARM-based laptop with thunderbolt/USB4 except for that one fruit company, but the first company to sell a linux-compatible device with ARM that I can use with my thunderbolt dock can take my money!
At this point for everyday types of tasks, most processors are complete overkill anyway. Even if an ARM based Framework laptop didn’t outperform or even match the latest Intel i7 mobile chips, I’d still be extremely interested if it allowed long battery life and cool, quiet operation. If I need to do more intensive tasks, I’ll fire up my desktop computer. But a 13" laptop with decent performance, 15 hours of battery life, a decent screen for content consumption, and no fan is basically my ideal, general use laptop.
And because of that, my current, general use laptop is a MacBook Air M1. I’ve never had it feel any temp other than ambient, it has no fan and is totally silent, it has a decent screen, the battery usually lasts me two or three days of general use, there is no intake/outlet to worry about when setting the computer on my lap, the arm of a chair, or wherever. I purchased it well before I heard about the Framework laptop.
I LOVE the concept of the Framework, and I’ve complained for years about devices removing features, reducing repairability, etc. I put my money where my mouth has been, and purchased a base, DIY Framework laptop. I will use it to learn Linux and to do some general “tinkering.” So far it works great and I’m happy with it. As much as it pains me to say it, it won’t be replacing my M1 Air just yet. Unfortunately, the M1 Air is essentially a throw-away laptop. If some component on it dies in a couple years, I’m sure there will be no economical avenue to keep it going. If it wasn’t so perfect for my use case otherwise, I never would have bought it. If I didn’t already have the Air, I probably wouldn’t buy one. I’d just use the Framework laptop. But since I already have it, I’m going to keep it since it is working so well for my everyday use.
But if the M1 Air ever dies, I’ll switch to full use of the Framework. And if Framework has a laptop with similar attributes to what I described above by then, they’ll get my money again.
Even for high-performance stuff, I honestly just run that stuff in the cloud these days. My only high performance tasks are
- Blender (but aside from rendering the iGPU does this fine)
- Gaming (slowly becoming less of an issue because I mostly run older/indie titles)
- Tons of monitors (4-6) for productivity (which I use a Thunderbolt GPU for)
Everything else I either run in the cloud (AWS) or have been able to do since my Lenovo T400 with a Core 2 Duo. Battery life matters WAY more than raw power, since my job is mostly writing with software development/research/3d modeling/GIS on the side.