ARM64 Open Laptop Concept | Hacker News Reader

ARM64 Open Laptop Concept

wiki.vero-apparatus.com

84 pointsedward46 comments

Related: Novena laptop, desktop and standalone board

https://www.crowdsupply.com/kosagi/novena-open-laptop

http://www.kosagi.com/w/index.php?title=Novena_Main_Page

Related to that: http://www.bunniestudios.com I'd forgotten it was called Novena but remembered Jacob Applebaum saying something about a "bunny" laptop with open source hardware :)

At some point virtually everyone has to "trust" something (e.g. That component X wasn't diverted/modified between purchase and arrival). However a set of open source, "many eyeballs" laptops would be a big win for privacy IMO.

AMD's A1100 is supposed to be "around 2.5x faster than AMD’s current low-power Jaguar-based server chip (the Opteron X2150), while maintaining the same TDP."[1] The X2150 seems to operate within 11-25W.[2]

On paper, this sounds like a thermal envelope that could still work within a 12.5" laptop like the venerable X220, although noted that the A1100 doesn't actually come with on-board video. Also, the A1100 dev kits seems to be currently packaged only for mini-ATX. Will all this actually somehow fit in an X220 chassis?

[1] http://www.extremetech.com/computing/175583-amd-unveils-its-...

[2] http://www.amd.com/Documents/Kyoto2150_QRG.pdf

The X2150 seems to operate within 11-25W.

That's much higher than some x86 based chipsets. Some of new Intel's Core-M Broadwell mobile chips are supposed to come in a 4.5W TDP. [1] But 11W is in range of some other older laptops.

These guys may be better off targeting a gaming laptop based on an i7 to ensure adequate cooling for a server-grade processor in a laptop form factor.

[1] http://www.extremetech.com/computing/189373-intel-details-co...

And it's mostly bullshit. Those laptops will either overheat, if the OEMs trust Intel with the "4.5W TDP" claim, or will be much lower performance, or they will actually have a higher TDP.

http://www.extremetech.com/computing/192893-the-first-core-m...

Also people need to discern between actual reviews and "we've just copied all the info from Intel's PR" reviews. My link points to an actual review, yours to the latter.

Also this:

"the Yoga 3 Pro allows for bursts of up to -- 12W total system consumption --- however"

Extremetech are shit.

That "review" appears to be written by someone who doesn't have the actual product - it's just a compilation of other reviews.

Also see this "first impressions" review:

http://www.notebookcheck.net/First-Impressions-Lenovo-Yoga-3...

3.7W to 10.7W power usage when idle, and about 20W under load, depending on display brightness. So the SoC seems to use at least ~10W under load.

Edit:

I guess I have to retract my statement. The 15W TDP Haswell Macbooks also require >30W under load in their tests. Since power is measured externally, losses from various components add up: conversion losses in the power adapter and voltage converters, simultaneously charging the battery while emptying it, ...

Yes, this particular choice of hardware sounds strange to me too, this CPU is intended for low power servers, isn't it? There are tons of ARM system on chips in 1..5W power envelope and have built-in video, audio and USB.

Also sentences like "Hardware choices will opt for longer battery life rather than 3D graphics performance" confuse me. The chips that have high 3D performance also usually have very low power consumption. Albeit in this case it will be problematic to find any kind external graphics chip, I assume the CPU has a PCI-e bus and they will have to pick a discrete PC laptop GPU.

There's a further problem with ARM chips. Unlike x86, where there's some kind of a standard for peripherals and software (ie. the PC "standard"), all ARM SoCs have very diverse configurations. There's no standard pin layouts, no standard "chipsets", no standard BIOS or bootloaders. All IRQs are different, memory maps vary, and standards like ACPI for power management, etc do not really exist.

I really wish that there were ARM-based low power laptops with as much open source firmware/drivers/kernel as possible, but unfortunately, this project does not seem very realistic to me. I hope that future ARM-based devices will have a better ecosystem and standards for building a complete computer around the CPU without being married to the software and hardware of the particular SoC vendor.

disclosure: I work for an ARM SoC manufacturer

> There are tons of ARM system on chips in 1..5W power envelope and have built-in video, audio and USB.

This is ARM64, not ARM!

There are no ARM64 SoCs in 1..5W power envelope that have built-in video, audio and USB.

Yes, there are. Although not many yet. I've been involved in building one that currently ships.

ARM64 (ie. aarch64) is not that different from 32 bit ARM.

And I've been involved in writing ARM64 compilers and ARM64 kernels. Saying ARM64 is not so different than 32-bit ARM is understatement of the year.

Of course they're different but when it comes to area, power and performance, it's not a huge difference. It's an incremental improvement, not a quantum leap.

Dude it's a totally new ISA. The assembly might look similar but it's got a totally different register set, instruction set, and memory model. It's definitely a quantum leap.

A quantum leap is a change of an electron from one quantum state to another - thus a multiple of the smallest possible change.

I don't understand this digression at all. Yes, ARM64 chips have approximately the same performance envelope as ARM chips. How is this relevant to anything in this discussion? If anything, it's an argument against tablet-level SoCs. Tablet-level SoCs have poor performance, so they want a better chip?

I don't understand the comparisons to ARM world at all. These people want to make an open (to the extent possible) ARM64 laptop. The ARM64 market is extremely scarce. There are very few chips, and even fewer are for sale. If you're not a huge player, almost nothing is for sale (yet). Those chose the available chip that has acceptable performance and requires fewest binary blobs.

Is there any ARM64 (!!!) SoC of comparable performance that requires less than of fewer blobs than this chip, and is for sale? If yes, I want to see it. If not, I don't understand the point of this discussion. Note that I mean actual SoC for sale, not some kind of board. And by available for sale I mean available in small number today, and without having to sign away your soul to buy it.

An underpowered phone SOC with a closed GPU requiring binary blobs sure sounds like a great fit for a project whose goal is making an open laptop.

I've used multiple attempts at ARM-based laptops / two-in-ones. So far they've been all been painfully slow. If they're targeting a x220 chassis, the constraints are going to be completely different from anything a consumer ARM SOC is ever deployed on. The standard battery on that is 63Wh, and it used 35W TDP CPUs. If there's a chip that can get laptop performance rather than phone performance, it makes perfect sense to spend at least a little bit of that massive thermal and power budget on it.

Perhaps the higher power tablet chips (ie. closer to 5-10 Watts TDP) could be a nice alternative? FWIW, I have used desktop Linux on a ~2012 ARM tablet chip and it wasn't fast but still a decent experience.

I'd be more worried about finding any sort of decent graphics chip. The discrete laptop chips tend to be for gamers (ie. high power) because all Intel laptop CPUs have an integrated GPU in them anyway (for mainstream customers). And sourcing discrete GPUs in small numbers is going to be difficult. Add in a requirement for open source drivers and it's even more difficult.

With luck AMD eventually bundles one of these with a small integrated GPU. They have somewhat well working Open Source drivers already.

AFAIK, they are using one of the AMD chips that is designed for ARM SBSA.

I don't understand the comments in this thread at all. This is an ARM64 chip. All comparisons with ARM chips are pointless and all suggestions that they should have used the X or Y ARM chip instead are missing the point.

What are the advantages of ARM64 over 32bit ARM?

What do governments in China and Russia use for high-security computing? There seems to be a valuable un-addressed market, or back doors in hardware don't exist, or back doors in hardware have been too hard to find, even for governments with the resources to conduct a search.

I suspect it's some type of system where they're even able to inspect the BIOS.

Speaking of... Richard Stallman uses the Lemote Yeelong because even its BIOS is open source[1]. The specs are a bit behind, but for his purposes, it seems to work well.

[1] http://richard.stallman.usesthis.com/

I saw him speak last night, and he had what looked like a Thinkpad X60 -- http://shop.gluglug.org.uk/product/ibm-lenovo-thinkpad-x60-c...

If that's true, then he picked very well. The X60 is pretty close to being my all time favorite kit. It's easier to carry around and packs more kick than an older Atom (certainly with Linux). I can see why he would choose it.

I actually haven't heard of LibreBoot before. Seems like a very ambitious project, but fits perfectly with RMS' beliefs http://www.libreboot.org/

I currently use the Samsung Chromebook 2 with Ubuntu via Crouton, and it runs nicely. But it's not the same as running a full GNU+Linux system with free software and hardware, and the on-board 16 gigs of storage is rather limiting. I would be all over this if the motherboard were made available.

You can boot your own kernel directly to a GNU/Linux rootfs. If you need more space, high speed SD cards work quite nicely.

I've tried Chrubuntu and it doesn't work as well as Crouton. The wifi, specifically, is flaky. Have you had better luck with the Chromebook 2 specifically?

It's been a while since I've used it and it turns out it's a first generation Samsung Chromebook (XE303). I don't remember having any WiFi-related issues.

The source code used to build the kernel for ChromeOS should be available so I'm surprised this sort of thing is a problem.

I think we should hack together an open serviceable laptop like Google Project Ara but for laptops.

So you can fix your screen, you can upgrade your memory you can upgrade the motherboard and cpu change storage, upgrade to never wifi standards when they are available.

Most Thinkpad T/X/W models (well, ones that haven't been turned into cheap Macbook clones by Lenovo) do this. Same for older Dell Latitudes and IIRC some HPs.

The machine I'm writing from (a Thinkpad X220) has had its' screen changed to an IPS, the HDD replaced with an SSD, they keyboard replaced to one with a different layout (GB->US), the RAM increased to 16G, the battery replaced to a 9-cell, a 3G modem added and 2x USB3.0 ports added via a ExpressCard module. All done based on a service manual from Lenovo.

High-end and professional laptops used to be very servicable and upgradeable, until the recent trend of form-over-function became popular (I'm looking at you, Macbooks and Intel Ultrabooks).

UEFI over the opensource Coreboot?

I think it is similar to what ARM SBSA uses.

These guys are ridiculously misguided.

UEFI, and not coreboot which is used in all of the arm chromebooks. A server chip in a laptop at 12W.

Yawn. Great idea but I'm not holding my breath.

Coreboot can't boot any ARM64 chip. i7-2640M (one chip in the original x220) has a TDP of 35W.

Tianocore EDK2, which they plan on using, is the reference UEFI implementation and is BSD-licensed. ARM actually develops and maintains the ARM-specific parts of the implementation.

Some people (e.g. authors of [1]) consider the UEFI reference implementation to be of poor quality. There was an exploitable (for privilege escalation from typical userspace administrator to system management mode) integer overflow there, and it wasn't a particularly tricky one[1].

[1] https://www.mitre.org/sites/default/files/publications/14-22...

Looks like there might be videos of some of the team presenting at this weekend's mini DebConf put online in the coming days: https://wiki.debconf.org/wiki/Miniconf-UK/2014#Saturday_sess...

Of note: "[Vero Apparatus] is a small group of people who are working on what they think is an under-provisioned niche in the computer market: full-power ARM-powered laptops. Daniel will be presenting their plans."

How about Nvidia's Denver chip? The single-core performance will really help web-browsing and I bet the power is very good for long battery life. ASUS zenbooks have a great chassis. My 2c.

Isn't the x220 a tablet?

http://shop.lenovo.com/us/en/laptops/thinkpad/x-series-table...

Seems an odd choice. Not something I'd want to own :/

EDIT: nm. there's a non-tablet version, just no page on main site: http://support.lenovo.com/us/en/documents/pd015812

X-series tablet models have a 'T' at the end.

Also, the X220T is not a "tablet" in the modern sense, it's a convertible laptop with a touchscreen.

I get the desire for DIY but why not just adopt an existing low cost hardware device and hack at it until it is extremely well documented?

Commercial hardware always becomes dirt cheap compared to small production custom stuff.

This month for the holidays there are $10 arm based smartphones that are pretty darn powerful (dual core 1.2ghz). No DIY project could ever match that.

There's very little ARM hardware in a laptop form factor in general, none of it is ARM64, and the ones that exist are mostly chromebooks, so maybe not the best build quality.

Hate to nitpick, but; spell-check.

I'd totally try this!

Reply on news.ycombinator.com