Linux compatibility for the ARM CPU is the easy part. Qualcomm makes most of the CPUs in android phones (which is linux-based already). This will almost certainly be done by launch day or shortly after for the X Elite.
The hard part is linux support for everything else in the laptop. It’ll be up to the team building the specific laptop (framework in our case). They’ll have to choose parts with working linux drivers or write the drivers themselves, and then they’ll have to test it all works together end-to-end and submit bug reports and patches to the linux project as issues arise. Same as they do for their intel and AMD laptops.
Unlike the x86 CPUs, an ARM CPU contains almost everything that usually is placed on a CPU mainboard. So official support for an ARM CPU usually means that almost the whole machine will work OOTB. Much less modular though, probably only the SSD and the memory can be replaceable, with the memory still questionable.
I think, creating an ARM option will be unavoidable at most within the next 1-2 years, due to thermals and battery life. If competitors tick the fan-less mode box, but the Framework keeps summing next to it, for more money, then it will be a pretty unattractive alternative.
But I imagine it can be difficult, to add the plug-n-play port support to such a SOC.
Also, will it reduce the sense of a Framework laptop? Upgradability will include to change the whole mainboard if I want to add RAM or more internal storage? Or will the new Snapdragon X support M.2 SSD slot?
According to the qualcomm website, the Snapdragon X Elite supports (among other things):
Internal 5G, WiFi 7 and Bluetooth 5.4
Up to three 4k displays (one internal at 120 Hz, two external at 60 Hz over DisplayPort 1.4)
NVMe SSDs over PCIe 4.0
M.2 E key modules (WiFi, 5G, Bluetooth…) over PCIe 3.0
3x USB 4.0, 2x USB 3.2, 1x eUSB 2.0
I couldn’t find if the USB ports support DP alt mode or if a mainboard would require dedicated DisplayPort 1.4 ports but all the prerequisites are there.
Edit: not sure about RAM but I assume that once LOCAMM2 becomes more available, it would be a good option.
Oh I would love too see a mobo upgrade for a FW13. I think that makes a lot of sense and there are plenty of linux users that would put it through it paces.
Just some wild speculation that I should probably keep in my head instead of sharing, but:
The current job posting for an electrical engineer specifically calls out experience with “recent SoCs, DRAM, and WiFi modules”. I want to believe they’re eluding to the Snapdragon X Elite, CAMM2 and wifi 7
This would be great option to have. Rumors suggest AMD is working on an ARM apu nicknamed Sound Wave, so based on the existing close relationship between Framework and AMD, that could be the more likely possibility. Either would be welcome in my book.
If the Snapdragon Elite X is as good as the previews and benchmarks suggest, and if Microsoft make Windows-on-Arm work as well as macOS-on-Apple-silicon, then I would love my next mainboard for my FW16 to be Arm-based. Maybe the Elite X2 next year…?
Snapdragon X Elite is a definitely mind changer for me. I would directly buy a FW13 with X Elite in it and install Linux.
It would be probably best Linux laptop with ARM arch. I was hoping to buy a lenovo for their good linux compatibility but Framework would be direct winner of this new area and will dominate (especially in enterprise IMHO).
Imagine you can upgrade your M1 macbook air with M2 or M3 processor. This is what FW13 can become.
With the Core Ultra series, Intel appears to have maximized innovation (via a deeply laborious shift to chiplet design + applying to the iGPU lessons learned from BattleMage) while still being deeply constrained by both its manufacturing process (7nm nodes) and the x64 architecture.
This raises questions about where laptop CPUs are heading. Snapdragon X strongly suggests that once one manufacturer shifts to 4nm ARM, the rest simply can’t compete regardless of surrounding innovation effort.
My question is: does Snapdragon X prove that the entire current competition from Intel and AMD is a waste of time until they both also shift to ARM on laptops?
How many wasted generations until Intel and AMD are both forced to fold towards ARM? Are we currently in some kind of wasteful limbo period where Intel and AMD are chugging out laptop CPUs that are doomed to being woefully obsolete the moment Snapdragon X forces everyone to switch to ARM?
Is Framework, the company, thinking about this question?
I wasn’t able to find any Snapdragon laptop review/benchmark (excluding some first party Geekbench scores). Can you provide some references/links for your considerations?
Reading stuff I’ve got the idea that good arm performance (current example is Apple) are due to software and better/more-expensive processor design rather than CPU instruction set.
ARM is a more efficient architecture inherently. It may be doomed to a slightly lower IPC in the long-run, but my crystal ball says better parallelism and thermal properties will ultimately outweigh that.
So IMO, yes ARM (and just RISC architectures in general) are where the market will be heading in the long term. And yes, the best consumer CPUs on the market right now are ARM. Also yes, we’re stuck with x86 from framework for now.
That said though, framework has already made it clear that they are interested in ARM (and even RISC-V) architectures. So I doubt framework is far behind microsoft right now. I’m sure they’ve got close eyes on the industry.
Arm isn’t inherently more efficient than x86. Many of the reasons Apple silicon has been so efficient comes down to things that are not tied to ISA. Things like favoring a wide design over clock speed, memory near or on package (Intel is going this direction for lunar lake iirc), a very wide memory bus, and always launching with a fab process advantage.
Most of these things are very expensive, but not necessarily tied to arm.
Snapdragon hasn’t proven anything at all yet, since the only data out so far is pre launch benchmarks shared by Qualcomm/MS themselves - never trust vendor data, no matter who it is. Until independent reviewers have access to an actual production unit, it isn’t real.
It’s a smaller ISA. At the end of the day, it’s less silicon to keep powered for the same job. Of course there are a whole lot of factors at play and you can definitely find low-power x86 devices and high-power ARM devices, but with all else being equal RISC architectures do have an efficiency advantage over CISC. It’s why we’ve historically seen RISC in mobile and IoT despite CISC dominating everything else.
Jim Keller has done interviews where he basically says thats so small it doesn’t matter, and if x86 (and arm) have a problem it’s legacy instructions for compatibility; I’d be inclined to trust him on this particular subject:
“When RISC first came out, x86 was half microcode. So if you look at the die, half the chip is a ROM, or maybe a third or something. And the RISC guys could say that there is no ROM on a RISC chip, so we get more performance. But now the ROM is so small, you can’t find it. Actually, the adder is so small, you can hardly find it? What limits computer performance today is predictability, and the two big ones are instruction/branch predictability, and data locality.”
Tldr is he’d choose riscV over arm and x86 because it’s newer with less legacy, and because it’s open, not necessarily because it’s inherently more efficient.