I own a MacBook Air M1. It’s awesome, fast and reliable all these years since they launched back in 2020. But I want a 13” Framework so badly, but I want a laptop with the flexibility of the Framework on ARM technology, it’s so hard to make? There’s no demand for it? It’s on the oven? 
But it has to be un expensive like the MacBook M1
1 Like
One of the reason the M1 is so good at what it does is that it is completely integrated (power savings, cost savings, more compact so more space for battery and all that). Also apple put trillion dollar company levels of engineering into the chip itself, that is unfortunately not something you can just easily replicate.
I really do wish there was something like that around but there currently really isn’t.
3 Likes
(personal opinion)
Only the Snapdragon 8 Gen 4 with Nuvia cores will be faster than Apple M2 core
https://twitter.com/Tech_Reve/status/1639770694631444480
So I expect an ARM Framework by 2024 or later.
2 Likes
That’s what I mean, if I understand correctly the ARM technology is the one we have in cellphones, why is so hard to make a laptop, I mean by an independent company like Framework? I totally would get one, no fans (that means is going to last more).
Its difficult to scale the performance of current ARM cores to the level expected of current x86 processors, and its even more difficult to get software to run on ARM. It is an entirely different instruction set, so codebases at minimum have to be entirely recompiled and often take significant reworks. Windows on Arm is still in its infancy.
The ARM in the M1 is a very different beast than what we got in cellphones.
2 Likes
at this point in time an arm processor with FW will only be feasible with Linux. the microsoft surface with arm processors is a buggy mess
1 Like
ARM is not what made the m1+ macbooks good. The only reason they chose ARM was that it was the only reasonable option at the time (and still is). x86 is restricted by patents and RISC-V not mature enough.
Some reasons why the M1 is actually so good:
- generally excellent engineering from apple with billions of dollars of R&D
- Apple’s vertical integration of software and hardware allowing for specific optimizations (flexibility is usually inefficient, e.g. the swappable RAM on frameworks is less efficient than soldered ones)
- The on chip RAM (hence why it’s completely inaccessible and initially was so limited)
- The massive jump in the manufacturing process (intel 14nm → TSMC 5nm)
And likely other reasons. ARM has no inherent benefit.
I think ARM is unlikely to ever succeed outside of apple. Considering how slowly the industry is moving on it, RISC-V will probably get there first, if at all.
7 Likes
God I hope high performance desktop riscv happens.
One of the reasons high performance arm isn’t that much of a thing is you’d need apple levels of lawyers to actually find out what license you need to be actually allowed to do it and to actually get it.
From recent interviews tenstorrent has a bunch of ex apple m1 engeneers and they went with riscv for their cpu architectures because arm would not let them do what they want. (They are mostly focused on ai stuff though)
1 Like
From where I’m sitting, ARM is too fractured, whereas I have been watching the RISC-V ecosystem take one massive leap after the other, all the while being way more open (with serious upstreaming efforts - eg. just look at the VisionFive2 pieces ending up in the upstream linux kernel currently.)
Try figuring out the boot process of ARM devices, it’s a nightmare, and usually you can’t anyway because you’re locked out, eg. if your phone doesn’t get updates anymore, you can’t do anything useful with them… (Then again, we don’t get coreboot on FWs either, and why is that part even locked in the first place, why can’t I replace the signing keys like I can replace secure boot keys… (okay, getting too far off topic, sorry…)).
3 Likes
Well, you could wait for Meteor Lake, supposedly is really power efficient, or even Arrow Lake, if Intel’s claims are true.
I am waiting for the release of the Meteor Lake Intel mainboards and the reviews to decide between the current upcoming AMD mainboard.
Well, I didn’t knew about that other technology, the RISC-V. As a regular consumer, not even a gamer (I like consoles over computers for that) all I want is reasonable price, not overheating or fans, fast and reliable (that means I prefer Linux over Windows) and powerful computer in case I need to do heavyweight lifting like video editing, image editing, audio editing or programming. All that has been covered by the MacBook M1, if you show me a computer on another system, I don’t care if is not on ARM, but if can make all that without fans, that part is super important and I think I’m not alone here, it’s extremely annoying to have a stove over your legs or hands, those days are over with the new MacBooks, if you offer me that. Take my money. Today Linux is so advanced, is getting to the point where it can emulate or has software so competitive that I can switch from the traditional mac or windows software. I think Framework needs to remember Linux for this big leap (if ever is going to take it). Maybe in that operative system is where they can avoid all that legal stuff and bring a game changer for the ones we hope the right to repair win.
I used an M1 Max 16" before and it has fans? It does let the internal temperatures reach rather high before the fans turn on though.
If this is what you want, why not get an M1 Macbook and install Asahi Linux?
I own a MacBook M1, but I want have more options, own another computer outside the “Apple garden”. And specially one I can repair, maybe that’s my urge that Framework can advance in the race against Apple, if people like me have a competitive option against Apple, because Windows is not competitive enough on the contrary, Windows is buggy and waste overall, if we have a competitive option all this personal computer panorama can change towards the user.
Makes sense to have a low power “non x86” cpu optionfor Framework, but a decent ideal chipset will likely not be available until at least a couple years. There is now RiscV chipsets for consumers at Rpi level. My guess is in a couple years there might be something that makes sense for a laptop. But right now there is no great option. You either go with a Snapdragon, and those are expensive and can give problem with drivers or there is nothing really that good.
1 Like
ARM even in servers are a walled garden frankly. ARM seems to attempt to make it hard for people who port over to return back to x86. Snapdragons are kinda the best but the closed source nature unlike RPi makes it difficult. Driver updates are also highly dependent on the vendor too.
I am interested in RISC-V but mainstream adoption (if ever), would be in quite a few years.
We would probably not see that same level of ARM performance as Apple as Banana has pointed out. Also, the ARM of Apple silicon builds in many features from the security core, to even the storage controller (which is why you don’t see a controller on the mainboard unlike NVMe drives etc). That insane amount of tightly integration between software and hardware will only be proprietary.
Technically you could have a slightly more open firmware then you desolder and reball storage/RAM chip modules for the firmware to recognise but there’s probably at most a limited list of modules that it accepts because it impacts RAM timings etc.
So probably no, in fact, looking at AMD now and things like Intel really low power chips, they aren’t as hot or as power hungry as you would think they are. Notice that the talk about ARM came during a time when Intel and x86 stagnated. And now with competition, x86 innovate further and less talk about ARM now. RISC-V is really interesting but I hope it doesn’t break compatibility (technically modern Intel and AMD x86 CPUs are not fully CISC either), so cutting off old code in RISC-V, how much really better in real life remains to be really seen.
That may happen quite suddenly and there are a lot of risc-v options in the market, arm seems to have really really pissed off a lot of their customers lately while risc-v continued to get better and lets them do whatever they want without a bunch of lawsuits between.
On the small scale risc-v is doing great, nvidia has risc-v on every gpu they make, wd on every harddrive and cheap risc-v micro controllers are showing up everywhere.
Yeah that integration has it’s pros and cons, it trades modularity, and a ton (like an unholy ammount) of engineering time for cheap production cost and great performance and efficiency for their intended tasks. You can really only pull that kind of stuff if you produce a huge number of the things.
x86 is bloated as hell, sure you may be able to get similar power efficiency to arm chips if you use the fanciest manufacturing nodes you can find and use way more die area for the same thing.
I really hope one of those server risc-v startups accidentally make an amazing core they can sell for desktop/laptop use. I could even see apple pivoting to risc-v for their M series if arm continues to sue everyone trying to do anything cool with their stuff, those guys really don’t do sunk cost.
1 Like