Not all vendors' Arm-powered kit is created equally, benchmark fan finds
- Reference: 1670329811
- News link: https://www.theregister.co.uk/2022/12/06/arm64_laptops/
- Source link:
Compared to x86 machines, which have the widest range of interchangeable, modular hardware in the history of computing, devices such as phones and tablets are almost polar opposites. A fondleslab is a relatively sealed unit, with a known combination of main processor, graphics processor, various networking interfaces and so on, designed and built to run a single OS that is customized for that hardware. If you're lucky, several versions of that single OS.
When it comes to the inexpensive highly integrated Arm device you are probably holding in mid air, the problems start with booting: there's usually no standardized firmware able to load multiple unknown OSes in the typical ad hoc PC way. There's no need if you're building it to run just one specific payload, meaning that, from the vendor's point of view, it's cheaper to do without.
[1]
Trying to get a general-purpose OS to load up on this and drive its hardware is not a trivial task, which is why [2]the Armbian project exists . For example, while [3]OpenBSD 7.1 boasted support for the Apple M1 and [4]version 7.2 added the M2 , this shows that definitions of "support" vary quite widely. OpenBSD [5]dropped support for Bluetooth completely in 2014, and for the typical uses of the OS, a flat unaccelerated framebuffer is good enough.
[6]
[7]
Various posts online are revealing some of how Apple's M1 kit works, the performance gap between M1 and M2, and how some rivals' kit compares.
Berlin-based developer Christian Adam has published an exhaustive [8]blog post examining the issues around running Windows 11 on a 64-bit Arm machine, the Samsung Galaxy Book Go 5G.
[9]
In the end, Adam bought three different laptops, plus a commercial driver manager and various other bits of hardware, and paid for someone to replace the LCD screen of one of them, then did another screen replacement on a second laptop himself. He managed to download the Windows installation image from a Microsoft Surface Pro X – running Arm hardware means no ready-to-run ISO images.
In the end, his verdict was:
It has a plastic feel to it, not very solid, seems more like a toy. The keyboard is not that great.
As for the last machine's performance:
It's like half of the CPU performance of an Apple M1.
And in order to get one fully working machine to use as his main computer, as he put it:
All in all I've spent €1,281, for which I could have bought an Apple MacBook Air Midnight, M2 - 8 Core CPU / 8 Core GPU, 8GB RAM, 256GB SSD priced at €1,274, but I wouldn't have had so much fun.
We suspect that not many people would be prepared to go to such lengths, but it does explain the levels of interest in Apple's Arm-instruction-set machines. The M1 and M2 family of in-house designed M1 SoCs set new performance standards for personal computers – so long as you're happy to run Apple's own macOS, of course. The Reg has covered [10]the user experience of Asahi Linux , as well as [11]the difficulty of reverse engineering its GPU .
[12]Microsoft ships non-Surface PC: a cheap Arm box for devs
[13]Google reveals another experimental operating system: KataOS
[14]Microsoft unveils native Arm64 support in the .NET Framework
[15]Arm confirms IPO delay till later in 2023, blames global markets
Asahi continues to make progress and improve its device drivers. Most recently, project lead Hector "Marcan" Martin [16]announced support for USB 3.0 devices, and better power management, but not yet for MacBooks' onboard speakers:
I decided to take one for the team and run some tests on my MacBook Air M2, and even with some sensible volume limits I quickly managed to blow up my tweeters. Oops!
Graphics driver developer [17]"Asahi Lina" also described her adventures reverse-engineering GPU drivers in a lengthy [18]post , which contains a lot of exclamation marks and more emojis than we're used to seeing in such technical matters, but is nonetheless fascinating. The early stages used a kernel driver which, although it sounds implausible, was partially written in Python. A more complete driver is being implemented in Rust, which she praises highly – although not without some [19]reservations . We've looked at [20]the difficulties of incorporating Rust into the kernel and the [21]modest scope of the initial support in the forthcoming kernel 6.1 . As Lina put it:
The kernel driver will not be heading upstream for many months.
We suspect that this could mean more complete support in the mainline kernel may not appear until version 6.2 or even later.
While the performance gap between Apple's M1 silicon and competing products such as the Qualcomm Snapdragon 8cx Gen 2 that Christian Land tested remains substantial, an interesting Geekbench benchmark [22]score was noted on Twitter by [23]ShrimpApplePro .
This appears to show the performance of one of [24]the forthcoming M2 Max models . It's a relatively modest step up from the released M2 hardware… but then again, some of the main designers of the M1 already left Apple, including [25]Gerard Williams to Nuvia (later [26]acquired by Qualcomm ), and subsequently [27]Jeff Wilcox to Intel , and [28]Mike Filippo to Microsoft . It is possible that Apple may yet face some [29]stiff new competition . ®
Get our [30]Tech Resources
[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/personaltech&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2Y491NoPmE@Hu0hkvQBzBiwAAAMQ&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[2] https://www.theregister.com/2022/03/03/armbian_project_releases_version_2202/
[3] https://www.theregister.com/2022/04/22/openbsd_71_released_including_apple/
[4] https://www.theregister.com/2022/10/21/openbsd_72_released/
[5] https://marc.info/?l=openbsd-cvs&m=140511572108715&w=2
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/personaltech&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Y491NoPmE@Hu0hkvQBzBiwAAAMQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/personaltech&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Y491NoPmE@Hu0hkvQBzBiwAAAMQ&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[8] https://cristianadam.eu/20221126/windows-arm64-samsung-galaxy-book-go-5g/
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/personaltech&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Y491NoPmE@Hu0hkvQBzBiwAAAMQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[10] https://www.theregister.com/2022/03/22/asahi_hands_on/
[11] https://www.theregister.com/2022/05/10/apple_m1_mesa/
[12] https://www.theregister.com/2022/10/25/windows_dev_kit_2023/
[13] https://www.theregister.com/2022/10/17/google_kata_os/
[14] https://www.theregister.com/2022/08/10/dotnet_481/
[15] https://www.theregister.com/2022/11/21/arm_ipo_delayed/
[16] https://asahilinux.org/2022/11/november-2022-report/
[17] https://github.com/asahilina
[18] https://asahilinux.org/2022/11/tales-of-the-m1-gpu/
[19] https://twitter.com/LinaAsahi/status/1570119306461204481
[20] https://www.theregister.com/2022/09/16/rust_in_the_linux_kernel/
[21] https://www.theregister.com/2022/10/05/rust_kernel_pull_request_pulled/
[22] https://browser.geekbench.com/v5/cpu/18972971
[23] https://twitter.com/VNchocoTaco/status/1597882566258098176
[24] https://www.theregister.com/2022/04/16/apple_m2_chips/
[25] https://www.theregister.com/2019/12/09/apple_nuvia_ceo_suit/
[26] https://www.theregister.com/2021/03/16/qualcomm_closes_nuvia_announces_chipset/
[27] https://www.theregister.com/2022/01/06/apple_wilcox_intel/
[28] https://www.theregister.com/2022/01/13/microsoft_apple_chip_expert/
[29] https://www.theregister.com/2022/05/02/qualcomm_laptops_nuvia/
[30] https://whitepapers.theregister.com/
Re: We need more modular systems
That was the case around 20 years ago but the creation of SOC's has made that IMHO, a thing of the past.
Once upon a time, you had a choice of 'Computer Fairs' to go to on a Sunday. I still use a pair of Dell Keyboards that I bought at the one in Addiscombe for £5.00 each.
The SOC and the levels of performance that Apple has shown with the M1 and M2, has shown us all how to do it. Performance without the heat (and fan noise) plus a long battery life.
Qualcomm and the rest will be racing to catch up.
I got my M1 16in MBP a couple of weeks ago. My workflows have been halved in time despite the new machine having half the RAM as its predecessor (2020 16in MBP). The old device will be passed onto my grandson once I get the arrow keys fixed under Applecare.
As far as I'm concerned, Intel and AMD are dead men walking. I will be moving my Linux servers to ARM as soon as I can get a small form factor system. I am using one R-Pi but it is a bit of a kludge. I'm fed up with the slow boot times of the SD card. Time to allow booting from M2 SSD's.
A meal is a selection of menu items.
This is not uncommon.
ARM is basically a menu of IP that you gather into your SOC with your own glue logic and IP.
As with all designs, you can make choices in this, trading speed, power, size and functionality.
Apples and pears
A desktop PC is a box with a motherboard and all sorts of bits can be added and removed and swapped around with bits from all sorts of manufacturers. So this need for interchangeable parts goes right back to the early days.
Pretty much no ARM device has ever needed to have any sort of equivalent. The closest is perhaps the Acorn RiscPC that accepted standard memory modules and you could change the processor card. But the ecosystem is for customised SoCs which is a fixed entity aimed at a specific device (or class of devices in the case of things like Snapdragon). There's no need for interchangeable parts and with integrated GPU and I/O and all, it wouldn't be possible anyway.
So mixing and matching on ARM hardware isn't really possible, but it's not right to compare against a PC, it isn't one. It evolved differently.
Re: Apples and pears
[Author here]
Yes and no. I think that in terms of cause-and-effect you are putting the cart before the horse here.
There is nothing inherent about Arm _per se_ that means that they have to be highly-integrated. It would be perfectly possible to design an Arm box with generic PCI-e slots and an offboard GPU, disk controller and all the normal gubbins of a PC, and if you did that, it could run Linux and Arm-compiled drivers for all the bits.
But everyone is paying attention to Arm again because the M1 chips are both so bally fast, and run so cool, that they make for very powerful desktops and also for exceptionally thin laptops with exceptionally long battery life.
I owned an Acorn Archimedes in the 1980s. This is not news to me. :-) My £800 ARM2 box outperformed the fastest £10,000 PC my employers had as a demo model. (Nobody much bought that model, partly because it was ten thousand quid. Excluding keyboard, monitor and DOS, as was IBM's wont.)
Now, Apple has put the Arm ISA back in that position. It is just about the fastest desktop or laptop money can buy.
*But* the thing is that Apple has achieved that by extremely tight integration and most of a computer on a single SOC: CPU, chipset, GPU, all the RAM, all in one package. And packaging that allows 2 SoCs to function as conjoined twins, too.
It is the integration that has permitted the performance, but the ISA does not demand such integration.
Re: Apples and pears
Apple has put the Arm ISA back in that position. (That is, of outclassing allcomers.)
Trouble is, A and M series chips are not available on the merchant market, and nobody else yet has anything close — not even with the ARM ISA. [1]Much [2]publicised [3]defections of Apple architects and engineers are touted as allowing chip vendors to catch up, but the best I've seen promised is the [4]end of 2023 — and that's probably predicated on Arm losing a [5]licensing dispute in which it seems to have a strong case.
Also, the literature about reverse-engineering the M1 shows that Apple has put up as many road-blocks as it can by patenting its speed-ups from here to kingdom come — although this is a game everyboy plays.
In short, don't hold your breath.
[1] https://www.theregister.com/2022/01/06/apple_wilcox_intel/
[2] https://www.theregister.com/2022/01/13/microsoft_apple_chip_expert/
[3] https://www.theregister.com/2021/11/16/qualcomm_pc_chip/
[4] https://www.theregister.com/2022/05/02/qualcomm_laptops_nuvia/
[5] https://www.theregister.com/2022/08/31/arm_sues_qualcomm/
Using Python for the driver
While it might not be the best tool for the job, modern machines are fast enough for you to use it to get some kind of prototype quickly than can later be written in C or Rust.
Really to get wider adoption....
We need more modular systems, upgradable cpu and memory seem to be the last great hurdle for this but that would also give less freedom to CPU makers since they'd need to build for everyone rather than tailor to a specific use case.
It's all abit awkward really.