Stories
Slash Boxes
Comments

SoylentNews is people

SoylentNews is powered by your submissions, so send in your scoop. Only 18 submissions in the queue.
posted by janrinok on Monday August 01 2016, @05:23PM   Printer-friendly
from the over-to-you! dept.

The goal of the EOMA (Embedded Open Modular Architecture) project is to introduce the idea of being ethically responsible about both the ecological and the financial resources required to design, manufacture, acquire and maintain our personal computing devices. The EOMA68 standard is a freely-accessible, royalty-free, unencumbered hardware standard formulated and tested over the last five years around the ultra-simple philosophy of "just plug it in: it will work".

With devices built following this standard, one can upgrade the CPU-card (consisting of CPU, RAM and some local storage) of a device while keeping the same housing (e.g. laptop). One can also use the CPU-card in different devices (e.g. unplug CPU-card from laptop, plug into desktop); or use a replaced/discarded CPU-card from a laptop for NAS storage or a micro-server. There are housings currently available for a laptop (can be 3D-printed in full, or in part to replace parts that break) and a micro-desktop; and there are plans for others like routers or tablets in the future.

There are multiple articles talking about this project and analyzing the hardware, for example from ThinkPenguin, CNXSoft or EngadgetNG. There is also a recent live-streamed video introducing the project.


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 1, Insightful) by Anonymous Coward on Monday August 01 2016, @06:01PM

    by Anonymous Coward on Monday August 01 2016, @06:01PM (#382718)

    As I posted on slashdot, the device while neat is under spec'ed for modern desktop usage. It has too little ram and too little storage space. It would be better suited as a alternative that is completely open to the raspberry pi and similar.

    • (Score: 1, Interesting) by Anonymous Coward on Monday August 01 2016, @06:06PM

      by Anonymous Coward on Monday August 01 2016, @06:06PM (#382722)

      That is how these small projects start off. No economy of scale, and takes time to get better specs. It is an ideological project meant to give users control over their own hardware. If you're looking for a product to compete with the slave labor factories of Asia, then you're doing it wrong.

      • (Score: 0) by Anonymous Coward on Tuesday August 02 2016, @12:47AM

        by Anonymous Coward on Tuesday August 02 2016, @12:47AM (#382903)

        Even as an ideological project it fails. All arm cpus in the past decade more or less have had a trusted execution core that is not accessible by the rest of the system. In phones this is the drm cpu that makes sure the media you play on it is legit.

        • (Score: 0) by Anonymous Coward on Tuesday August 02 2016, @02:48AM

          by Anonymous Coward on Tuesday August 02 2016, @02:48AM (#382964)

          Is that the case with this one? Why would the FSF seriously consider endorsing this, then?

          • (Score: 0) by Anonymous Coward on Wednesday August 03 2016, @06:32AM

            by Anonymous Coward on Wednesday August 03 2016, @06:32AM (#383509)

            The FSF wouldn't.

            And the campaign would refund your money if you pledge for a Libre Tea computer card and no RYF certification is obtained.

            https://www.fsf.org/resources/hw/endorsement/criteria [fsf.org]

            No Spying or back doors

            Please be aware that we can't check products for spy features or back doors, but if we find out about any we will drop our endorsement unless they are promptly removed.

            Further, because spyware or a back door may take many forms, we will need to work out precise criteria as we go along. The criteria stated below are our first attempts, and we may change them if we come across examples which they handle wrong.

            "Spyware" is software that transmits or makes available to anyone other than the owner any information about the owner's or the product's identity, location or activity except (1) when the owner asks to do so, (2) when inevitable due to communication protocols used, or (3) when required by law. In the case 3 applies, the product should come with prominent warning, plus a URL for where to obtain more information.

            A "back door" is any functionality that gives remote control over any aspect of the device operation or its software. It is normal for computers to have facilities for updating software or for letting people log in remotely. These become back doors when they are enabled by default.

        • (Score: 1) by lkcl on Tuesday August 02 2016, @07:11PM

          by lkcl (6308) on Tuesday August 02 2016, @07:11PM (#383265)

          Even as an ideological project it fails. All arm cpus in the past decade
          more or less have had a trusted execution core that is not accessible
          by the rest of the system. In phones this is the drm cpu that
          makes sure the media you play on it is legit.

          everything that you've said, whilst true for *specific* ARM SoCs from
          manufacturers who *specifically* work hand-in-hand with the cartels
          and U.S. Government spying agencies, for this *SPECIFIC* SoC that
          has been chosen, everything that you've said is completely false.

          i would not work on a project for five years to bring you an RYF
          Certifiable device if i had *any* indication that the A20 was designed
          to spy on people.

          you can get a *small* glimpse into the evaluation process that has
          been ongoing for five years, here:
          https://www.crowdsupply.com/eoma68/micro-desktop/updates/picking-a-processor [crowdsupply.com]

      • (Score: 2) by frojack on Tuesday August 02 2016, @04:47AM

        by frojack (1554) on Tuesday August 02 2016, @04:47AM (#382996) Journal

        Byfield addressed this very issue in a recent blog. Why Free Hardware Fails [linux-magazine.com]

        He pretty much agrees with your assessment.

        --
        No, you are mistaken. I've always had this sig.
    • (Score: 1) by mafm on Monday August 01 2016, @06:17PM

      by mafm (6305) on Monday August 01 2016, @06:17PM (#382732) Homepage

      It's true that the specs of the current CPU-card are probably not enough for desktop usage for many people. Some people who don't use many applications or browser tabs at the same time might be happy with the specs for desktops, or they might like to have an extremely light laptop that they can 3D-print and repair for years if parts break.

      I think that it's good for micro-servers now and for years to come, or routers or devices like tablets, if the ecosystem thrives and such housings are created (there are plans about that).

      Also, there are plans to look into ARM 64-bit CPUs (and other architectures) for future CPU-cards, some work has already been done in that regard.

    • (Score: 2, Informative) by lkcl on Tuesday August 02 2016, @07:20PM

      by lkcl (6308) on Tuesday August 02 2016, @07:20PM (#383267)

      As I posted on slashdot, the device while neat is under spec'ed for modern desktop usage. It has too little ram and too little storage space. It would be better suited as a alternative that is completely open to the raspberry pi and similar.

      no it would not, and you're missing the point entirely of the exercise.

      first point: let's reference the conversation - https://news.slashdot.org/comments.pl?sid=9462323&cid=52617167 [slashdot.org]
      summary: to even remotely consider entering the over-saturated SBC market would be totally pointless. i have better things to do with my time, because i have the ability to predict issues and have the guts to stand up and solve them even if i don't immediately know how.

      second point: modular upgradeable architectures are, by definition, upgradeable. by backing the project now, you're supporting us *being able* to bring upgrades in a continous, affordable and ethical way. i've found two SoC candidates that i'm investigating: one's the R8 (so we could bring people a $20 computer card) and the other's the Samsung Octa-core A64 which we could do a massive upgrade in about a year's time, instantly quelling all of the complaints.

      third point: if you want to carry on using a backdoored processor (from intel or AMD), please feel free to continue to explore that space. for everybody else: aside from X200s which are regularly failing in the field due to their age and being over-stressed by modern OSes and uses, there really isn't anything else. we've looked.

      fourth point: comparing this project to mass-volume priced well-established incumbents is not going to help you... or this project. i'm seeing this mistake being made enough times now that i'm going to do a special update about it. you can't possibly seriously compare a crowd-funded project with a MOQ of 250 units and an ethical business model to *any* profit-maximising unethical multi-billion-dollar company like apple, HP or Dell and expect me to take your anonymous comments seriously.

      does that clarify matters for you?

  • (Score: 3, Interesting) by Gravis on Monday August 01 2016, @06:14PM

    by Gravis (4596) on Monday August 01 2016, @06:14PM (#382729)

    the problem with the card design is that they don't utilize high speed IO, the video uses most of the connections and you have no dedicated IO for storage. oh and it gets worse. the underpowered SoC is from AllWinner, a flagrant GPL violator. i want a libre laptop too but this isn't the answer.

    • (Score: 1) by mafm on Monday August 01 2016, @06:35PM

      by mafm (6305) on Monday August 01 2016, @06:35PM (#382736) Homepage

      There are two updates discussing the questions raised here:

      I think that I read in one of the updates or explanations around that Allwinner is one of the few ones which allow to work with a small number of orders (the goal is 250 at the moment).

    • (Score: 0) by Anonymous Coward on Monday August 01 2016, @10:15PM

      by Anonymous Coward on Monday August 01 2016, @10:15PM (#382859)

      The point of libre computing isn't to be the best in a technical sense. If anything, improving on that would come later. And this is libre, even if AllWinner violated the GPL in the past.

      • (Score: 1) by lkcl on Wednesday August 03 2016, @07:48PM

        by lkcl (6308) on Wednesday August 03 2016, @07:48PM (#383733)

        exactly. judging someone by their past is a bit like hitting an animal for something it doesn't understand. we need to be able to do psychological "instant reward / gratification". they've made their money: they don't care. boycotting them isn't going to help *us* get Libre-compliant hardware. real simple, i wish there were more people who understood this.

    • (Score: 1) by lkcl on Tuesday August 02 2016, @07:30PM

      by lkcl (6308) on Tuesday August 02 2016, @07:30PM (#383272)

      the problem with the card design is that they don't utilize high speed IO, the video uses most of the connections and you have no dedicated IO for storage. oh and it gets worse. the underpowered SoC is from AllWinner, a flagrant GPL violator. i want a libre laptop too but this isn't the answer.

      we see these misunderstanding being made time and time again, gravis. i'll try to be brief as i have a lot to cover, apologies in advance

      there are 2 USB ports, one is USB2 the other is up to USB 3.1. there's also SD/MMC on the interface and a 2nd one on the end of the computer card (just this particular one - other Computer Cards will have their own I/O). so that takes care of the I/O misunderstanding

      the video is RGB/TTL 18 pin, that's only 21 out of 68 pins - the standard had 5 years of refinement, i think that's quite a long time to do a decent analysis: if you feel i've missed something do contact me on the mailing list after reviewing the whitepaper http://rhombus-tech.net/whitepapers/ecocomputing_07sep2015/ [rhombus-tech.net]

      the HDMI output on the EOMA68-A20 is a 2nd output so you can actually run two screens simultaneously. we feel that's probably enough for most people. so that deals with that misunderstanding.

      the SoC is the best that can ethically be obtained - there's an update about it, please read that first, if you know of any other SoCs that can fit the ethical criteria do let me know, i need to evaluate them https://www.crowdsupply.com/eoma68/micro-desktop/updates/picking-a-processor [crowdsupply.com] so that deals with that misunderstanding

      the situation with allwinner is complex, there are some extremely powerful investors each of whom has carved out their own niche, we're working with allwinner to provide a financial argument along the lines of, "these people are buying this processor because it's GPL compliant, they want to place an order for 50k units of this other processor and will buy 1 million later but you have to fix the GPL violations FIRST".... **NOW** we've got their attention. if you boycott THE ENTIRE COMPANY guess what happens? no conversation, no fixing the problems. so boycotting allwinner just fucks everybody over. don't do it. so, that mistake / misunderstanding clarified.

      lastly, this is a modular upgradeable architecture - which you've never seen before so there isn't anything you can compare it against. so the misunderstanding about "this system is too slow therefore don't get it" is perfectly understandable... but wrong. see the other post "titled clarification" above for full answers on this.

      sorry, bit of a rush, apologies for the curt tone, i'm dealing with somewhere around 25 different forums at any one time.

    • (Score: 0) by Anonymous Coward on Wednesday August 03 2016, @06:54AM

      by Anonymous Coward on Wednesday August 03 2016, @06:54AM (#383512)

      There's an option that is a passthrough card. When you plug this into the laptop enclosure then it isn't a laptop but a screen and keyboard and trackpad with battery that you can plug to your mobile phone or TV stick or development card or even laptop to get a second screen, with USB and HDMI cables,

      You can support the concept, get a repairable laptop with your choice of processor and alow the project to carry on to produce better cpu cards in the future so that you can remove the cables and the external computer or mobile.

      You don't have to buy an Allwinner to support the project.

      I don't know about the future behaviour of Allwinner and how such a company can be impressed by 200 or 300 (or 1000 if it comes to it) CPU cards sold. Or even is such a company can control itself even if impressed. But I know there are no good SOCs to choose.
      In order to change that there are 2 ways:

      - Dump lots of money to buy or build hardware manufacturers (possibly SOCs and others). Hopefully it works with RISC-V or similar, but takes money and time.

      - Accept that the performance we got some years ago is as good now as it was then and stop being addicted to the latest features because this is what feeds the oligoolies that keep abusing consumers to the point of spying, remote controlling or even remotely bricking your computer when they decide you should buy another.
      http://boingboing.net/2016/04/05/google-reaches-into-customers.html [boingboing.net]

      No current CPU is decent enough to use. If you don't have your own fab, you can either use second hand computers while they don't fall apart, don't use computers at all, or,
      as a mitigation, not really a solution, use as few CPUs as you can to feed your enemies the less you can. So if you can use the CPU in your mobile for you laptop use cases,
      get an EOMA68 laptop with a passthrough card, and you get one less CPU sold by this abusive industry.

  • (Score: 2) by oaklandwatch on Monday August 01 2016, @06:33PM

    by oaklandwatch (5011) on Monday August 01 2016, @06:33PM (#382735)

    Rhombus Tech just made actual EOMA68 devices available through their new crowdfunding campaign.

    https://www.crowdsupply.com/eoma68/micro-desktop/ [crowdsupply.com]

    There's a lot of good information about their philosophy -- they're hoping to get certified by the Free Software Foundation. But all ideology aside, I just think it'd be cool to have a wooden micro-housing unit for my CPU.

    • (Score: 0) by Anonymous Coward on Monday August 01 2016, @07:02PM

      by Anonymous Coward on Monday August 01 2016, @07:02PM (#382751)

      Allwinners get certified by the FSF? Given how poisoned that well is, I don't see that as very likely.

      • (Score: 0) by Anonymous Coward on Monday August 01 2016, @08:02PM

        by Anonymous Coward on Monday August 01 2016, @08:02PM (#382788)

        They think there's a chance since they'll be avoiding the MALI drivers entirely, or not even wring the pins or something. They're explanation was that they should get the endorsement if it's not even possible for users to accidentally use the offending component.

        • (Score: 1) by lkcl on Tuesday August 02 2016, @07:44PM

          by lkcl (6308) on Tuesday August 02 2016, @07:44PM (#383283)

          They think there's a chance since they'll be avoiding the MALI drivers entirely, or not even wring the pins or something. They're explanation was that they should get the endorsement if it's not even possible for users to accidentally use the offending component.

          that's a very close approximation, yes, which needs very little additional clarification to be an accurate summary. substitute "The *FSF's* position" for the word "They", "Their" for the words "They are" which were shortened to "They're" ... remove the 2nd use of the word "even", and you need to substitute "average completely non-technical point-and-click-only end-user" for the word "users"... a few other things and you're good to go. so let's do that:

          The FSF's position is that there's a chance since they'll be leaving out the MALI kernel drivers entirely, such that the memory-mapped MALI hardware is totally undetectable from userspace. Their explanation was that they should get the endorsement if it's not possible for average completely non-technical point-and-click-only end-users to accidentally use the offending component, or even be able to tell that it exists.

          further clarification is that as far as the FSF is concerned, technically-minded end-users can look after themselves, and can make well-informed decisions. technically-minded end-users are capable of discerning the difference between "free" and "non-free". it's *everyone else* that the FSF's position takes care of.

      • (Score: 1) by lkcl on Tuesday August 02 2016, @07:36PM

        by lkcl (6308) on Tuesday August 02 2016, @07:36PM (#383275)

        Allwinners get certified by the FSF? Given how poisoned that well is, I don't see that as very likely.

        the FSF does not pass judgement based on past actions: that would be foolish and severely counter-productive. they're interested in *right now*. we would not be applying for RYF Certification if the A20 was not entirely GPL compliant right down to the bedrock.

        the way to solve the problem is to find a *financial* incentive. if we can say "YOU WILL LOSE MONEY IF YOU DON'T GIVE US THE SOURCE CODE", and dangle a deal for 1 million units in front of them, guess what happens? it actually turns out that they can be convinced even for the small amount of 50,000 units.

        but if you start being a dick and saying "it's poisoned, fucking boycott the fuckers" guess how that's going to turn out, eh?

        am really getting tired of hearing people telling me that they're boycotting allwinner when i'm working REALLY HARD to solve the problem.

  • (Score: 0) by Anonymous Coward on Monday August 01 2016, @06:49PM

    by Anonymous Coward on Monday August 01 2016, @06:49PM (#382740)

    this is a start. hopefully it gets funded.

  • (Score: 2) by choose another one on Monday August 01 2016, @06:57PM

    by choose another one (515) Subscriber Badge on Monday August 01 2016, @06:57PM (#382745)

    Any one else scratching their heads wondering why there is a PFY option but no BOFH option, or is it just me?

    • (Score: 1) by lkcl on Tuesday August 02 2016, @07:47PM

      by lkcl (6308) on Tuesday August 02 2016, @07:47PM (#383287)

      Any one else scratching their heads wondering why there is a PFY option but no BOFH option, or is it just me?

      uhnnn... you don't mean any of these, surely... :) http://www.acronymfinder.com/BOFH.html [acronymfinder.com]

    • (Score: 0) by Anonymous Coward on Tuesday August 02 2016, @10:06PM

      by Anonymous Coward on Tuesday August 02 2016, @10:06PM (#383367)

      If you mean print-it-yourself, that's there. It's $450 and labeled "PIY".

    • (Score: 0) by Anonymous Coward on Wednesday August 03 2016, @07:05AM

      by Anonymous Coward on Wednesday August 03 2016, @07:05AM (#383513)

      There's some cultural mismatch. I bet the parent was refering to a series of stories published in a sensationalist tech newspaper, theRegister, and maybe elsewhere which had among others 2 characters:
      BOFH : Bastard operator from hell (IT sysadmin).
      PFY: Pimply faced youth (apprentice)
      http://www.urbandictionary.com/define.php?term=PFY [urbandictionary.com]
      http://www.urbandictionary.com/define.php?term=BOFH [urbandictionary.com]

  • (Score: 2) by LoRdTAW on Monday August 01 2016, @10:08PM

    by LoRdTAW (3755) on Monday August 01 2016, @10:08PM (#382857) Journal

    Ethical? If anything, this stupid design will make the hardware even more disposable.

    Can I upgrade the RAM? No. It's soldered on.
    Can I upgrade the storage? No. It's soldered on.

    What exactly is the benefit here? And what happens when you need more RAM? What happens when your eMMC goes bad like what happened to my beagle board? Now you have a useless card with working RAM and CPU but broken storage. A single point of failure is all you invested in. Same goes if you RAM/CPU goes bad or the card fails to boot. How do you remove your eMMC chip and recover your files? Now you need a whole new card and the existing one is trash along with your data. No thank you. Been there, done that.

    How is this better than a proper laptop? That I can move the card to another case? Big deal. I can plug a 27" monitor into my laptop as well as keyboard and mouse without unplugging its motherboard or CPU. Now I have a desktop! Plus I can remove/replace the disk, and upgrade the memory. Two birds, one stone.

    There is a limit to how much you can upgrade a system. It's great that you want to save the environment and all but the method you introduced is absurd. You want a true enviro firendly computer design? Look no further than the humble ATX standard. I can swap any component I please to upgrade is small steps. An SSD here, new GPU there, and I'm done. I have put modern AMD and Intel boards and GPU's into ATX cases that are 10+ years old. My favorite is a Sun Ultra 20 (originally Opteron based) and an SGI 330. How's that for recycling?

    • (Score: 1) by Chrontius on Monday August 01 2016, @10:41PM

      by Chrontius (5246) on Monday August 01 2016, @10:41PM (#382869)

      Don't forget that the µSD slot is there for mass storage.

    • (Score: 4, Interesting) by Foobar Bazbot on Tuesday August 02 2016, @12:21AM

      by Foobar Bazbot (37) on Tuesday August 02 2016, @12:21AM (#382891) Journal

      First, let me be clear -- I'm going to talk as though this standard has succeeded already, and there's multiple CPU cards and multiple systems to use them in; after all, the whole point of this is to bootstrap the EOMA68 ecosystem, and if it only results in a single A20 CPU card, a laptop, and a minidesktop/breakout, it hasn't succeeded at all. So i won't address issues like "But there's no point in modularity if there's only one module that fits!". Those aren't invalid criticisms, but they're inevitable at this point -- and you'll have to judge for yourself the risk it never gets past this point.

      Anyway, you're coming at this from a desktop perspective, and comparing it to the most desktop-like laptops. Note that quite a few laptops do have soldered-down RAM. Storage is arguably a downside, but even there, a few of the netbook/chromebook type laptops that this actually competes with have soldered-down SSDs as well. Your perspective is absolutely not wrong, but it's not the most useful to understand what EOMA68 could achieve compared to its direct competition.

      ATX is great for desktops, but it doesn't do a thing to help with laptops, tablets, and embedded devices -- and that's what the EOMA68 standard is about. It doesn't really extend down to mobile phones, though a brick-phone with EOMA68 is just conceivable, and I personally would love it. It also only covers part of the laptop region, specifically netbook/chromebook type things, though the eventual type-III PCMCIA-based version, with higher power limit, will stretch that range. The reason it's currently focused on laptop and mini-desktop is partly historical chance (a KDE-based tablet was actually planned as the first mass-produced chassis, but fell through), and partly their usefulness for development on other platforms. (Plug a CPU card into a mini-desktop for development, plug it back into your new tablet chassis for testing, repeat till it works.)

      I mentioned "embedded" -- to be clear what sort of thing I mean, you know the trend of people taking appliances that would normally be connected to a PC (especially 3D printers, but other stuff too) and sticking a Raspberry Pi or similar with a small display on to make them standalone? EOMA68 is great for this sort of thing, because you get choice of processors easily (compare adapting a Raspberry Pi hat into a BBB cape), and potentially "free" upgrades with hand-me-down CPU cards. And if you're making a project with a custom PCB, it's relatively easy to add an EOMA68 slot -- the CPU card has all the difficult and/or multilayer stuff done, so you can add it to a simple double-sided PCB. Or you can make your own projects designed around the EOMA68 from scratch -- it's no harder than designing a "hat", "cape", or such for other single-board computers, but you gain futureproofing and choice.

      The EOMA68 actually helps a little with hardware failures (after all, compared to a tablet with everything soldered, at least you don't also have an LCD, touchscreen, battery, etc. rendered useless by your eMMC failure), but that's just gravy; the main point (from the consumer perspective) is to beat the upgrade treadmill. When the next hot processor comes out, and everyone bloats up their code to require it, you pull the still-working card from your laptop chassis and put a new one in. Then you put the "old" card in some other system (e.g. digital picture frame, router, 3D printer), replacing its even older CPU, and it gets a "free" upgrade. Or to put it another way, the cost of the new CPU card is amortized over two, three, or even more devices that receive upgrades, whether directly or second-hand.

      • (Score: 1) by mafm on Tuesday August 02 2016, @06:15PM

        by mafm (6305) on Tuesday August 02 2016, @06:15PM (#383240) Homepage

        Thanks for your comment, it really highlights the strengths and key points that the project tries to address.

        Yes, desktops are still quite modular, but the devices that have cropped up in the last few years do not work in that way, and as soon as some component fails or lags behind (e.g. memory of a tablet) the device is as good as dead. I have several of these gathering dust around the house.

    • (Score: 1) by lkcl on Tuesday August 02 2016, @08:01PM

      by lkcl (6308) on Tuesday August 02 2016, @08:01PM (#383292)

      Can I upgrade the RAM? No. It's soldered on.

      the RAM ICs (only 2 of them) cost $3 each. it would be financially unviable to consider socketing them (SODIMM sockets are around $1.50), and they're maxed out anyway.

      Can I upgrade the storage? No. It's soldered on.

      the NAND IC costs $5 and is only one of the storage options. again it would financially unviable to consider socketing them. other storage options include *TWO* micro-sd slots and *THREE* full-speed USB2 sockets. storage isn't going to be a problem.

      What exactly is the benefit here?

      the benefits are huge... and not specifically the ones that you're envisioning, because it sounds like you're used to high-end desktop PCs. the whitepaper "scenarios" section goes over a *SMALL* list of the benefits. http://rhombus-tech.net/whitepapers/ecocomputing_07sep2015/ [rhombus-tech.net]

      And what happens when you need more RAM?

      upgrade with a $30 to $60 computer card. keep the old one until you're confident that all apps and data have been transferred and are fully operational. don't tell me that you can do that with a standard laptop or desktop PC.

      What happens when your eMMC goes bad like what happened to my beagle board?

      i'm assuming you have backups. in the case of the A20, which is "unbrickable", it boots from external MicroSD as a first priority (not the NAND). sounds to me though like you were running ext4 and probably swap. generally a bad idea. you should look up the "read-only rootfs" scripts for debian - it's doable, and there's one for remounting read-only and one for read-write so you can do OS upgrades etc. without destroying the NAND / eMMC. the general idea is you run /var/log and pretty much everything else off of shm / tmpfs. it works really well, and speeds up your system in the process.

      Now you have a useless card with working RAM and CPU but broken storage. A single point of failure is all you invested in.

      not with an A20 you don't. i deliberately picked the A20 because of the "unbrickable" design.

      if your $5 processor breaks or your $6 worth of RAM, my advice is: don't worry about it. put in the *spare* computer card which you bought for that exact eventuality (or dust off the previous generation one that you kept around) and carry on working.

      you're imagining that there are problems where there aren't any. imagine the same scenario with a standard laptop. you ever tried disassembling a mass-produced laptop? you seen the warnings about the apple 2012 macbooks? they state VERY CLEARLY, be VERY CAREFUL so as not to puncture the battery and cause a lithium fire.

      by total contrast with an EOMA68 design just push the button, pop out the faulty computer card, put in the spare... you're up and running within under a minute.

    • (Score: 0) by Anonymous Coward on Tuesday August 02 2016, @09:10PM

      by Anonymous Coward on Tuesday August 02 2016, @09:10PM (#383341)

      One of the main points of this design is to literally replace the whole motherboard while keeping the rest of the machine.
      Less waste, way cheaper in the long run, user-friendly, etc.

      Sure that means less flexibility for RAM upgrade (and other individual parts) at first,
      but who knows, maybe a more expensive version with sockets could be a possibility
      in the future. But it has to start somewhere.
      Worst case, that's a comparatively small compromise since a more powerful card would still be relatively cheap.

      That ATX thing looks like a good idea, I' don't know about it. But I'm pretty sure it can hardly be fully libre.
      Probably less portable/adaptable, and costs way more. It also can't be as ethical, as eco-friendly,
      can't be repaired as easily (and for less money), etc.

      Seriously, if you still think it's a stupid design, you're still focusing on the finger pointing away to the moon, so to speak.
      You focus so much on performance right now (in its infancy) when it can only get better with time and support.
      Performance will come.

      • (Score: 0) by Anonymous Coward on Wednesday August 03 2016, @07:52AM

        by Anonymous Coward on Wednesday August 03 2016, @07:52AM (#383529)

        ATX is just the format for motherboards inside "normal" desktop PCs. A kind of standard that defines board dimensions, connectors, etc. so that you can build a PC picking among different cases, power supplies, motherboards, etc. I don't think the CPU sockets are part of ATX or even the DIMM sockets, it's just something the motherboard designers often put in. CPU sockets change every generation of CPUs and DIMMs are only compatible to similarly fast buses, so it's not such a mix and match freely, but there's quite a lot of choice.

        The format itself is allright for the desktp use case. Its diminishing sales are related to that everyone already has desktop PCs and it's hard to justify new features or performance, and other problems:
        - common operating systems are poor and people resist to change to uncommon ones
        - people fall addict to mobility and prefer computers they can carry with them always
        - maintaining a PC is too difficult for some people that they prefer thowing out a device every year or two. That's partly because of the internet wars and malware business but also because the poor software choices and habits, not an intrinsically unavoidable problem.
        - simply out of fashion ?

        The realproblem is that all modern CPUs to put in those ATX motherboards are cracked from factory to allow third parties to remote control, spy and sabotage your computing if they (or whoever gets that access in the future) want to. They carry unreplaceable operating system/hypervisor and server software which controls your operating system. And the CPU won't even access RAM before this hypervisor is verified as unaltered (with a digital signature) and set up.
        The problem with older ATX motherboards is that they carry proprietary software in their BIOS ROMs that can still take control of the computer. Those BIOS can be replaced with libreboot but it takes a lot of reverser engineering, so there aren't many options already available.

        See https://libreboot.org/faq/#intel [libreboot.org] for details.

        Once the chips themselfs come cracked, you can only try to avoid the manufacturers that do so. But:

        - Manufacturers are few

        - The portion of CPU manufacturers that allow you to boot an operating system under your control keep decreasing. Those that still allow you are the ones selling less powerful processors so that the overhead of cryptography and system management mode is too much compared to the rest of the design. And those have other problems for 100% free software operation

        - Despite the moore law dubious keeping pace, and the use cases for more performane being more dubious, people keep massively buying more powerful but more remote conrolled devices because of network effects (they want to see videos with excessive resolution, etc.) and that it is to hard for lay people to analyse the threats. It's easier to just compare price, RAM size, number of cores and clock. This spec race means manufacturing equipment needs to be replaced often, and that makes it more difficult in capital to build processors, so helping to keep the market an oligopoly.

        I really would like that the most performant and cheap computers were the more free, but it will never ever be so. Your data, your freedom and the possibility to screw you when somebody pays for it are parts of the prices you pay for this powerful computers that you buy at little money cost and high non-money cost. If that's the deal you like, you're normal, most people prefer this, and I stopped caring whether they understand it or not. If yo want another deal then look at EOMA-68, tinkerphones, talos, ministry of freedom, think penguin, the Pyra handheld and others. Or in general, if that's too complex for you but don't want to be ripped off, go and look at RYF certified hardware. Of the ones I know EOMA is the most environmentally aware. But only if successful, of course.

        You'll find some ATX motherboards that are RYF. But very few. I think I remember 2 that you can buy online and use without any installation or reverse engineering. I may not be aware of all of them.

        • (Score: 0) by Anonymous Coward on Wednesday August 03 2016, @06:45PM

          by Anonymous Coward on Wednesday August 03 2016, @06:45PM (#383705)

          Thanks for the info.

          Ok so basically, the argument of ATX being as modular (if not more)
          is at least more meaningful than the laptop argument, since when the laptop is dead,
          most of it is.

          Indeed some powerful ATX motherboards are fully libre. But they cost an arm.
          But fully libre, comparatively very cheap, and with a bright future performance-wise, EOMA68 has this.

          Privacy matters, it's not for some lunatics: https://www.privacytools.io/ [privacytools.io]
          But it's more than that, it's about the ability to improve, build upon, and do whatever you want with your machine.
          No industry-imposed limits.

  • (Score: 2) by jmorris on Tuesday August 02 2016, @01:48AM

    by jmorris (4844) on Tuesday August 02 2016, @01:48AM (#382931)

    If you want a headless server, buy Pi. Everything but the video is open and the Pi3 is a heck of lot faster for less money.

    If you want to run a desktop, buy AMD/ATI if you need performance and Intel if you don't. None of the ARM processors has OpenGL support that I'm aware of. None, even if you can get the closed drivers to run on your kernel. EGL != OpenGL. Vulkan should change that in the future, but only if you can stand closed drivers and limited kernel options. Unaccelerated framebuffers suck. They suck harder when driven by underpowered ARM processors which are intended to have hardware acceleration for damned near everything, leaving them to mostly just coordinate the co-processors and try to get any processing that can't be accelerated done before the thermal throttle shuts them down.

    Now to slagging this particular product. Allwinner is actually a pretty reasonable choice, all things considered, because they do low quantity and do reveal everything they have the legal right to reveal, i.e. everything except Mali. The fact noname 'gookie monster' shops churn out junk with Allwinner chips and don't release their GPL required minimum isn't Allwinner's fault. The form factor of EOMA is dumb. Repurposing PCMCIA only sounded like a good idea until put into practice. Not enough pins. Gotta go with something denser. You need Displayport (gets multiple displays on minimal pins) or a pair of HDMI controllers. The product under discussion does have a mini-hdmi but it is on the side where you would have to rat nest cables and forget using a direct digital link to a laptop internal panel from it. No storage. 8GB onboard is too much for a boot partition yet far too small for a modern distro and seriously unfit for the ones coming in the next few years. And no, MicroSD doesn't cut it. Nice to have one, don't expect to run your primary OS from one. SATA. Not really optional unless it is a toy server, buy a Pi if that is all you want. USB2.0 could be forgiven if it wasn't the only bus available and having to run both storage and network across it.

    • (Score: 1) by lkcl on Tuesday August 02 2016, @08:05PM

      by lkcl (6308) on Tuesday August 02 2016, @08:05PM (#383294)

      apologies but i haven't got time to go over every misunderstanding you've made, here, it's too much: perhaps someone else can help out here. all of these things have been answered in some detail on other forums. including in answers here, above. many of them are answered in the updates as well as in the FAQ on the crowdfunding page https://www.crowdsupply.com/eoma68/micro-desktop/updates [crowdsupply.com]

    • (Score: 0) by Anonymous Coward on Tuesday August 02 2016, @09:57PM

      by Anonymous Coward on Tuesday August 02 2016, @09:57PM (#383361)

      Since lkcl cannot respond to this, I will do my best. If I left anything out, it's because it's outside of my knowledge zone.

      If you want a headless server, buy Pi. Everything but the video is open and the Pi3 is a heck of lot faster for less money.

      I stopped paying attention to the Raspberry Pi after the first one, but if the third one is like the first one, you need that proprietary GPU driver to even boot the thing, making it no better than literally any x86 computer in the market in that regard. (The only thing it's better at is in not having a known backdoor.)

      If you want to run a desktop, buy AMD/ATI if you need performance and Intel if you don't.

      All modern x86 processors (since 2010 for Intel, or since 2013 for AMD) mandate proprietary programs which can act as very powerful backdoors. More info here:

      https://libreboot.org/faq/#intelme [libreboot.org]
      https://libreboot.org/faq/#amdpsp [libreboot.org]

      Additionally, computers like the A20 computer card are much more energy-efficient than any x86 computer.

      None of the ARM processors has OpenGL support that I'm aware of.

      You can use Regal:

      https://github.com/p3/regal [github.com]

      There may be other compatibility layers available. I would also like to point out that the OpenPandora community has been able to adapt several OpenGL applications to work with OpenGL ES just fine; this is not hypothetical.

      Furthermore:

      • EOMA68 is just a standard, and architecture-independent. One of the other SoCs being considered right now is one with a MIPS processor, for example.
      • There isn't anything inherent about ARM that makes full OpenGL support impossible. It's just that GPUs on ARM SoCs tend to only support OpenGL ES at the hardware level. You hinted at this yourself.

      Unaccelerated framebuffers suck. They suck harder when driven by underpowered ARM processors which are intended to have hardware acceleration for damned near everything, leaving them to mostly just coordinate the co-processors and try to get any processing that can't be accelerated done before the thermal throttle shuts them down.

      I have personally used a much weaker processor without any hardware acceleration at all, on the OpenPandora. That wasn't even a problem for me. The problem for me on the Pandora, no matter how I used it, has always been its low RAM. I even had people tell me that I couldn't possibly have a decent experience running Tiled with a software implementation of OpenGL, and in reality I had no problems with it at all.

      The Libre Tea is not unaccelerated. It is only specifically 3-D that is unaccelerated. 2-D acceleration works, so it's mostly 3-D games, heavy desktops like KDE, and applications that work with 3-D objects (e.g. Blender) which are affected.

      It's not about the one processor anyway, though. If this project is successful, upgrades can be made available, at a really cheap price. That's the whole point.

      8GB onboard is too much for a boot partition yet far too small for a modern distro and seriously unfit for the ones coming in the next few years. And no, MicroSD doesn't cut it. Nice to have one, don't expect to run your primary OS from one.

      How does a microSD card not "cut it"? A microSD card today can be as much as 256 GB, depending on what you're willing to spend, and with access times which are easily fast enough. The only problem with microSD is that it's flash memory and has all the problems flash memory has (most importantly, it's destined to become useless eventually depending on how often you write to it), but NAND is flash memory, too.

      SATA. Not really optional unless it is a toy server, buy a Pi if that is all you want.

      Again, a Raspberry Pi is not an option: you need to run a proprietary program just to boot it. But specifically regarding SATA, lkcl has already explained that SATA was excluded from the standard because requiring SATA would severely limit options for SoCs to use.

    • (Score: 0) by Anonymous Coward on Tuesday August 02 2016, @10:09PM

      by Anonymous Coward on Tuesday August 02 2016, @10:09PM (#383369)

      Current intel and AMD processors can't be trusted. The system won't boot unless proprietary signed code is run that takes control of the computer, like a kind of hypervisor that can be remote controlled.

      The rapsberry Pi needed proprietary software to boot the GPU that then starts the CPU. I'm not current with all the variants now.

      The only 3D acceleration that was usable with 100% free software was intel, only available with intel CPUs with the problems I just hinted, and Nvidia (thanks to nouveau, not Nvidia).
      Now both required proprietary signed code to run. ATI GPUs need AtomBIOS which is proprietary code running on a kind of interpreter. There's Freedreno for Adreno but most chips
      with adreno have problems of the modem controlling the whole system and running proprietary software that can be remote controlled. Then there's Freescale Vivante, reverse engineered
      with etnaviv to some extend, ...

      Listing all the case is long. But the point is that we are running out of hardware that e can control. We're even runnign out of hadrware that we at least don't know that is remote controlled
      by the manufacturer, its government, the future owners of both, the defectors or attackers of all those.

      About PCMCIA I don't know enough. The features it allows seem a good compromise. Some other connectors are good for setting up a system but do not stand plugging and unplugging often.
      Some can't be sourced economically at small quantities. Number of pins and speed of signals is maybe a problem, but not one that hasn't been well considered and accepted for good reasons. Future EOMA standards might have more pins.

      Storage is always the more the better, so you're right. I don't remember if 8GB was because of cost, power, space or something. But as long as there's a microSD card that's not a problem. I'm wrting this form a desktop that boots and operates from a microSD card. It's not that slow.

      SATA and ethernet were in an older version of EOMA68 but were sacrifized to fit USB3, I think.

      I'm asleep I may add links another day...

  • (Score: 2) by Foobar Bazbot on Tuesday August 02 2016, @03:01AM

    by Foobar Bazbot (37) on Tuesday August 02 2016, @03:01AM (#382970) Journal

    In previous discussions about EOMA, I've spent a lot of time explaining how it appeals to different parties. This is important, because no matter how appealing something is to consumers, if manufacturers don't want it, it won't succeed, and vice versa. The whole point is to have an ecosystem with lots of CPU cards and lots of devices (of all sorts) to use them in -- and that means it has to be appealing to everyone involved. So here's an explanation of why I think this standard has a chance, because it does stand to benefit everyone.
    End user -- relatively non-technical
    It's all about future-proofing, upgrading, and flexibility. You upgrade the CPU card in one system (let's say a tablet or netbook), and not only do you not have to repurchase the rest of the system (e.g. screen, battery, mechanical chassis, keyboard) along with it, you also get to hand the previous CPU card down to the next-most-demanding application (maybe your spouse's tablet, maybe an old tablet with a lousy screen that lives on your nightstand as an alarm clock). Or, conversely, if you want a new tablet with a better screen, you can get it and move your old CPU card over, keeping your data and settings all right there, and not paying for the CPU card again.

    Or maybe you buy an EOMA68-compatible digital photo frame. (Are digital photo frames still a thing? Seemed everyone was buying them for their mothers a few years ago...) It has a couple USB ports, a screen, and the dumbest/cheapest EOMA68 card imaginable. Displays pictures off USB drives, and can't do much else. But when you upgrade your laptop, you end up with a spare "last year's" EOMA68 card -- plug that into your photo frame, plug a mouse and keyboard in the USB ports, and you've got a little desktop where visiting friends/family can check their email and such. (And of course, it keeps displaying your pictures as a screensaver.)

    Additionally, there's some pretty nifty market shifts that EOMA68 might just make happen. For instance, a dumb TV or monitor can be outfitted with an EOMA68 slot for very low cost, and sold as a "smart-ready" TV, or sold with a pre-installed CPU card (with video-streaming software) as a smart TV. I'm thinking the simplicity of having one hardware build for both smart and non-smart models, and the marketing gimmick of "smart-ready", could make this obsolete both dumb TVs and inflexible smart TVs altogether. (And as I'm not a fan of current "smart TVs", I'd love the ability to unplug their "smart" module, and plug in my own EOMA68 card running desktop Linux.)

    Swapping one card in and out of multiple devices may be too much hassle to matter for most folks -- with ubiquitous internet connection, it seems an easier sell to just synchronize data through the cloud rather than plugging a card and rebooting to make apps, data, and settings persist between e.g. laptop and tablet. I like this capability, and believe I'd use it, but I'm hardly a typical user.
    Hardware hacker
    Of course there's the simple stuff, where you use the EOMA68 CPU card and a minimal breakout board in much the same way people use single-board computers, to add standalone capability to peripherals (e.g. 3D printer) that would normally be connected to a computer. Again, what you gain is flexibilty, and "free" trickle-down upgrades.

    But what about bigger projects -- ever want to make a laptop that suits you just right -- this keyboard, that display, maybe you really like a certain hinge mechanism, etc.? The casework's not insurmountably hard, or maybe you can even modify an existing laptop, but the big hurdle is the motherboard. Any worthwhile CPU is a pretty big BGA package, and you probably need 4 layers just to route all the signals out from under it. A bunch of those signals are very high frequency, and that makes routing them over any distance more complicated. Almost all projects of this sort end up using a mass-produced single-board computer of some sort, such as a Raspberry Pi or BeagleBone. In a sense, an EOMA68 CPU card fills the same role -- it's has the BGA routing and high-bandwidth routing taken care of, and exposes a neat interface to connect your PCB to. But it beats the currently-popular options two ways. First, you're tied into one platform by your choice -- if you go with R.Pi, you can't swap it for a BBB later, without redoing your PCB. (You may be able to upgrade to a newer flavor of R.Pi, but that's a very limited selection.) Likewise if you choose EOMA68, you can't go with the other options, but the difference is, there's not just one line of EOMA68 cards. Anyone can make a new EOMA68 card, with whatever SoC they fancy, so you'll have a decent selection available. Second benefit, depending on the PCMCIA socket used and the thickness of CPU card used, is reduced thickness compared to the typical SBCs with board-stacking connectors.
    Manufacturer (of tablets/netbooks/etc.)
    Reduced fixed costs related to design and prototyping. When a hot new SoC turns the market upside down, making a tablet with last years processor becomes a losing proposition. So these manufacturers scramble to turn out a new board with the new SoC, incurring lots of design and prototyping costs. And if they're not fast enough, they may not sell enough to make those costs back before the next change leaves them playing catch up again. But with EOMA68, you can keep shipping the same tablet, whether you sell it sans CPU card, or just ship it with a different CPU card. Likewise, when you do redesign (maybe for a higher-resolution screen, or perhaps just to make the casework look more like the latest iPad), you have less design costs because you've left the hard bit (SoC and friends) for someone else.

    In addition, there are niches that go untapped now, because it is believed (often correctly!) that they won't generate enough sales to cover the fixed costs. For instance, I loved the Fujitsu U820; for those who don't know, it's a paperback-sized mini-laptop, convertible, 5-6" screen, and -- this is great -- a trackpoint-ish pointing device mounted where it's usable in tablet and laptop mode, so you can actually use the convertible mode without smudging the screen up. The screen was too dim for outdoors, the CPU was slow (800MHz Atom), and it had a resistive touchscreen (principally for use with the included stylus), but I'd honestly pay $500-$1000 for one with the CPU and screen of a Samsung Note 3 or newer (includes Wacom digitizer as well as capacitive touch, though I could do without the latter), and a normal desktop OS. This will never happen -- it's just not worth designing it for me and the few hundred others who love tiny laptops. But between the reduced cost of producing one run of these, and the increased utility to buyers (because it won't be obsolete in 2 years, you'll just put in new CPU cards) translating into higher prices, some company could turn a neat profit that just wasn't there before. (Or perhaps that niche is still too small to be profitable. Even if so, you can see how reducing fixed costs shifts the balance toward small batches to clean up on niche interests.)
    Manufacturer (of CPU cards)
    Volume, volume, volume. To make a new CPU card for some new SoC, you've got fixed costs similar to doing the board for a new tablet with the new SoC. But whereas you'll never be a particularly huge part of the market for tablets with a given SoC, it's entirely possible that the first manufacturer to market with a CPU card for that SoC will have the whole market. It'll be low-margin, but high volume. (It's also possible that SoC vendors will produce their own CPU cards -- after all, what better way to get people to buy your SoC than to package it as a complete drop-in module.)

    • (Score: 1) by mafm on Tuesday August 02 2016, @09:31PM

      by mafm (6305) on Tuesday August 02 2016, @09:31PM (#383351) Homepage

      Loved this comment as well, thanks for sharing your insights!

  • (Score: 2) by migz on Tuesday August 02 2016, @09:03AM

    by migz (1807) on Tuesday August 02 2016, @09:03AM (#383037)

    This would have made more sense years ago when computers were slow, but in a world with GBP 5 Raspbery Pi, we're just gonna throw the old one away.

    This might make sense as a way of providing extreme customization.

    Want a 10" laptop with a 100Ah battery with a config that draws 1A at max cycles? Then decide you want a bigger screen. The ecosystem needs to extend way beyond the "Tower" for this to work.

    Also a use for shared resources, and privacy nuts.

    • (Score: 0) by Anonymous Coward on Tuesday August 02 2016, @08:29PM

      by Anonymous Coward on Tuesday August 02 2016, @08:29PM (#383314)

      Look, we all want performance and a great price.

      But you miss some the core aspects :
      - eco-friendly
      - ethical
      - basically one ghost in many shells, and that "ghost" is upgradable without rendering the rest of the machine useless
      - fully libre, which isn't an advantage for privacy "nuts" only
      - 3D printable case, how cool is that? you can personalize colors, but also shapes

      And it's still somewhat affordable.

      We're talking about upgradable libre hardware, and you're thinking performance?
      The Pi doesn't even compare.

  • (Score: 1) by jman on Tuesday August 02 2016, @01:00PM

    by jman (6085) Subscriber Badge on Tuesday August 02 2016, @01:00PM (#383106) Homepage
    Reminds me of the first Apple-1 ( http://www.cnet.com/news/apple-1-wooden-computer-possibly-hand-built-by-steve-jobs-could-be-yours/ [cnet.com] ).

    If only their current products were as easy to work on!