so, the benefit delivered by your FSF certification is that I can pay for an ARM based machine on which I can run Linux, slooowly, without having to worry about driver incompatibility. Or, I could buy an old thinkpad and get the same benefit (things just work) by running Windows on it quickly. Or buy an old macbook and get the same benefit as with Windows, plus a UNIX based OS.
we get this reaction a lot. i take it you’re used to 20-150 watt intel processors? the cost of the lowest-end intel processor is i believe somewhere around the $30 mark, retail, which is around the same cost as the entire computer card @ MOQ 250 units. these SoCs are between $2.50 and $7 so it’s a totally different market.
i’ve answered the heat dissipation question a number of times, including on here - you should be able to find two answers (because some got deleted by the webmasters by mistake)
regarding processing power: yes that’s right… right now. and that’s fine. it’s enough to do word processing, internet, email, print some documents, and watch films. it’s the “Good Enough Computing” paradigm. oh, and you won’t get spied on by your hardware, either.
later there will be double the speed, double the ram, double the storage, for exactly the same amount. by the time 20nm and 14nm and below become popular amongst tablet-style processors, we’ll be able to take advantage of these and package them up so you can buy an upgrade for $50 instead of throwing away a $500 laptop. that’s what this is about - saving you money, long-term.
… but if you’re used to higher performance, you can always buy a higher performance laptop and be happy with that.
the use of a personal pronoun in association with “FSF Certification” is not correct. i do not “personally own” something called “an FSF Certification”
and in the future upgrade for around $50 instead of $500, indefinitely. yes.
[quote] Or, I could buy an old thinkpad and get the same benefit (things just work) by running Windows on it quickly.
so, you’re comparing a mass-produced 2nd-hand product against a crowd-funded concept that’s just getting off the ground, running a windows OS. there’s no common ground to make this a legitimate comparison.
to make a legitimate comparison, let’s imagine that the project was funded over 10 years ago. you would have been able to buy a mass-produced 15.6in Desktop Housing about 8 years ago for around $200. 5 years ago you would have been able to pick one up second-hand for say… $60. you could choose to buy a second-hand computer card for… $10 off of ebay, or you could buy a new one for around $30, with a quad or octal-core samsung processor. you’d definitely be able to buy a $50 windows card (the z8300 atom).
that would be a more accurate scenario to compare against the example that you gave. how does that sound?
again, same incompatible comparison. to make it a valid comparison you’d have to compare this 10 years into the future (or pretend that it was funded 10 years ago), so that there exists a huge ecosystem and a thriving second-hand market in EOMA68 products.
so i appreciate you bringing up this comparison as it allows me to provide a correct (more realistic) one, so that people can see the benefits in terms of cost-savings by being able to imagine buying second-hand EOMA68 compatible products.
if you’d like to help us get to that scenario, where we can bring you octa-core computer cards and beyond, that’s what this crowd-funding campaign is for.
crowd-funding is about backing ideas, not about setting up a shop and placing orders under “contract of sale”. crowd-funding is a gift economy. you gift us, we gift you.
on pretty much every forum except for those related directly to software libre, we see a really surprising amount of vehement over-reaction. these over-reactions are usually grounded in a comparison of this very early crowd-funding phase to “mass-volume” pricing levels. they’re really useful as a counterpoint.
the other thing i feel is more subtle, and i believe it’s this: people really do not like to be reminded that they’ve had to make some serious compromises, both in terms of cost, privacy and convenience, to keep up-to-date with their computing hardware.
that we are offering something at all - even if it we have to take several steps to get there because we’re not doing this with massive VC funding (which would be an ethical sell-out anyway…) - people get really upset and don’t clearly understand why, and it’s because we’re reminding them that they had to make compromises that they’re not really comfortable admitting.
ok, this is particularly fascinating (and a good example). @Glaurung you start off well, saying “please don’t compare apples to oranges”, but then by sentence 5 you make a direct comparison of an ethically-run MOQ 250 crowd-funded project with a mass-produced unethically-driven product!
… you can’t have it both ways, man
also, you’re forgetting that upgrading is as easy as pressing a button… and is possible. you can’t upgrade the whole computer inside a laptop case: they’re designed for obsolescence. some of them are even DANGEROUS to try to upgrade, such as the 2012 macbook pros. they take hours, require extreme care, require special tools, and if you puncture the battery (which is very easy to do on those 2012 macbooks) you’ll cause a lithium fire.
by contrast, i’m deliberately asking the very early backers to assemble their own laptops so that when it comes to repairing them they’ll know how they’re put together.
i’ve designed the machines so that “Right to Repair” groups such as the people at the https://therestartproject.org/ Registered Charity in London can work with people to help them out.
… taking a leaf out of the “Monster’s University” book (ok film if we’re going to be pedantic)… would you prefer that i stop helping out and empowering Charities such as The Restart Project? because that’s ultimately what you’re criticising, here, and it’s not sounding very nice… for you!
@Glaurung… ok believe i’ve given you enough opportunity, here, and thorough acting in a disingenous fashion you’re clearly out to cause trouble. i’ve been able to utilise your points for clarification (for other people’s benefit - not yours), so in that regard you’ve proved very useful and helpful. at the same time whilst you continue to degrade the conversation i think you’ve done enough harm to yourself so that continuing to interact with you is counterproductive, and, furthermore, that other people can safely ignore you as well.
true… but look at the example of the arduino, and also the fairphone (which i’m reluctant to mention because of the mistakes they’ve made, but their hearts are in the right place).
people are prepared to buy arduinos from the original manufacturers despite the higher prices instead of going to china where the clones are cheaper… because they love the idea of supporting the people who make them.
people buy the fairphone products because they know that some of the money is going towards ensuring that the factory workers get a pension.
… so the real question should be: what is it about the mass-produced products that make people so resentful of paying anything other than the absolute minimum amount they can possibly get away with for them?
it’s much worse than you might imagine it to be. the profit margins are around ten to twelve percent… for EVERBODY. as in, not 10-12% for the factory, 10-12% for the middle-man, 10-12% for the shop… 10-12% DIVIDED amongst everybody. this is one of the reasons why Dell, Lenovo and others do white-listing in the BIOS of their laptops for PCie WIFI modems, so that they can “Get You on the minor items”, by massively overcharging for replacement parts.
but this practice should really tell you everything that you need to know: that these companies are operating unethically and people know it. everyone knows that microsoft threatens the OEMs so that their cartel is maintained… but what choice do they feel that they have (if they keep on buying)?
by total contrast for the right kind of design we know that people will pay more. and, here, that’s where i have the advantage, because i actually want there to be an ecosystem based around EOMA68 where there are third party manufacturers (they will need to ensure that they are compliant, for safety reasons, so will need to receive a Certification of Standards Compliance from me)
but i am not out for maximising profits above all else. i have too much else to do, after EOMA68 is stabilised and in mass-volume production. i’ll keep an eye on it for the next decade and beyond, but i’m not here for the purposes of “milking it for all it’s worth” as many people would be.
you’re reaaaally going to have to qualify that and give me some references so i can investigate REAL fast what you’re talking about, and do an update on it. it would be preferable if you could point at real-world exploits from respected security researchers who have released full source code and CVE vulnerability reports.
the only references i can find googling “Trustzone A20” are here: http linux-sunxi.org Talk:TrustZone - certainly i’m not even enabling Trustzone in the kernels that i’m compiling.
more info needed - i’ll need to investigate and do an update. i’ll also post on the mailing list to ask if anyone knows about this.
update: ARGH this is a damn nuisance, being restricted on the number of replies and links, damn nuisance that the webmasters didn’t do their jobs properly and deleted my account on here sigh okay @mewyn you’ll need to reconstruct this link:
http: wiki.xen.org wiki Xen_ARM_with_Virtualization_Extensions Allwinner
could you explain to me (for everyone’s benefit) what’s the problems that you’ve encountered (with full links to publicly-verifiable material) with the above successful virtualisation - use of XEN - on the A20?
… but from what i can gather, that’s absolutely no different from running a standard linux kernel on a standard x86. nobody runs virtualisation on x86 just to be able to deal with buffer over-runs or other types of attacks: you use SE/Linux, and other techniques.
my feeling is that you’re jumping to conclusions. or, at least, not comparing like-with-like and then saying that ARM (and Allwinner) are worse for security when a feature that’s NOT NORMALLY USED happens not to even be enabled, but if it WAS it would be INEFFECTIVE… well… if it’s NOT USED then it DOESN’T MATTER
mmm… the person you know, if they’re not publicly disclosing their work and are remaining anonymous, are being highly irresponsible. if you can kick his butt on everbody’s behalf and get him to publish, other people can investigate and confirm the validity of his investigations. if neither his credentials nor claims can be validated (because he’s remaining anonymous and not publishing), we have absolutely no information on which to be able to make any kind of risk assessment.
ok so it’s a new day, chances are high that i can reply here instead of having to arse about editing previous replies and thus destroying the logical order of the conversation thread… sigh sometimes dealing with spam just ain’t worth the hassle… ARGH no i can’t… damnit. okay so it goes here instead…
so. the “like-for-like” situation is to not be using TrustZone - or any kind of Virtualisation - at all. on intel x86, nobody runs XEN Hypervisors or other virtualisation just to deal with buffer overruns, web browser script injection attacks (XSS) and so on: apart from anything the complexity as well as performance hit is too great. i did actually once set up XEN virtualisation on web browsers, used the guests to run applications and the host to run the X-Server. it was complex as hell. so if you’re actually running that type of setup (on x86) good for you!
so, when we’re comparing like-for-like, there may well be problems with the A20’s Virtualisation (which the person that you know has not disclosed yet, which is quite irresponsible of him), BUT… it’s not a technique that is used in everyday average deployment of OSes… so my assessment is, it’s a non-issue.
the normal way to deal with buffer overruns in web browsers for example would be to report the bug and have somebody investigate and fix the bug. the normal way to deal with XSS script injection is to not use the website, report the bug to the webmaster and have them fix it. everybody has to do this regardless of whether you’re running Virtualisation or not, whether you’re running x86, ARM or MIPS - it doesn’t matter. Virtualisation only contains the OS in which the applications are running from affecting other OSes running other applications on the same real hardware. it’s far too crude a tool, in other words.
now, if it was possible to do Virtualisation of individual applications that would be a different matter. even better would be to be able to do Virtualisation at the function call level. the Cambridge Capability System developed in the 1960s / 1970s did this: it applied the concept of “Capabilities” right down to the function parameter level. it was a research project… and it stayed a research project.
bottom line, your friend is researching something that i don’t think is relevant to the majority of use cases to which this $7 processor is put. it’s in exactly the same use-case-scenario that x86 is in, and nobody’s complaining loudly about that.
Baloney. If that was the case, you wouldn’t have to offer copies of the hardware as rewards to raise your funds.
I thought you said this wasn’t about ideological purity?
It’s very simple. Computers are tools. Computing power is a direct measure of their utility as tools. You are offering to create a tool that works more poorly (is slower, does not run as many OSes), but which has been blessed by the ethical purity of open source from top to bottom. At this time, you are offering to provide these poor but ideologically pure tools in kit form for people who place no value on their time, or in assembled form for people who prefer to actually use their time using their tools rather than putting them together.
I was not trying to be hostile. As someone who does not belong to your church, I was merely observing that your proposed tools offer very poor value for the dollar to those like me who do not care much about the religion of open source or its doctrines of ideological purity. For someone who cares about having a top to bottom open platform, i am sure that the cards are very tempting (the laptop, not so much).
To insist as you did above that you are not in fact peddling ideological purity, on the other hand, is the height of disingenuousness.
Again, since you didn’t bother to reply, why are you here complaining about this? Why is the very existence of this project soooooo offensive that you need to double down on repeated negative and nasty comments on it?
That sounds somewhat pragmatic, but tools also have different uses, and I’d argue that their utility or efficiency is best measured against this use rather than put forth as one universal objective metric. Are you simply saying that it is a poor match for your uses and concerns, or are you hoping to be sufficiently general here as to speak for others?
Might it also not be disingenuous to suggest that “value for the dollar” has its own ethical and/or ideological baggage? Such as the management of the dollar itself, and the corporatist mass-production scalings of consumer culture? I am not speaking for or against those things, but I need to acknowledge that they are not ideologically value-neutral.
Besides, one of their selling points you gloss over is a lack of hardware backdoors, which a commodity laptop running Windows won’t get you. If that’s not what you need, try to not cry too hard about it.
The other thing is Allwinner SoCs are terrible for the paranoid. They are fine for hacking around with, but if you want to truly ensure that there’s not been any software tucked away somewhere by someone; you can’t do that with this chip. The TrustZone implementation is easily broken and the datasheets lie about the presence of a SMMU. This would be fine if you can be assured that only code you vet ever runs on the system; but as soon as you start using the web all bets are off. It’s pretty trivial to exploit and get a rootkit hidden on a linux system with an Allwinner processor.
If I were to have a paranoid machine, I’d probably go with a Chromebook and verify and sign my root image and get that running within the TrustZone. That way I could at least vet my software and be assured that it won’t be tampered with.
If you’re not using TrustZone you’re fucked. You can never verify that something hasn’t rooted your system without using it; that’s what TrustZone is there for. As far as sources, it’s “I know a guy”; I don’t have anything more that I can say besides Allwinner SoCs are the worst if you want to be able to guarantee privacy and security.
ok so thank you to everyone for the comments: we’re coming to the end of the topic’s open comments period, if anybody would like to get in touch and keep track of the project, the mailing list is here: http://lists.phcomp.co.uk/pipermail/arm-netbook/
Do you think anything will actually make them approach this productively? I doubt it based on the history of Internet discussion boards. They’re here to complain about other people doing or making something that they don’t want and/or like. Why? Who the hell knows?
You’ve yet to reply with why the mere existence of this project bothers you soooooo much. Cat got your tongue? Unable to form a coherent response? Hence the “butthurt” remarks. You’ve got nothing but being unhappy that other people are doing something that you don’t like that you weren’t going to buy anyway.