A64 OLinuXino OSHW Linux Laptop idea becomes more real :)


Few weeks ago I blogged about the idea to make OSHW Laptop based on Allwinner A64 64-bit SoC.

Today we received the first samples of the laptop plastic body.

The quality of the plastic parts is very good!

As you can see we have already sourced the plastic body, the battery, LCD display, keyboard, touchpad, speakers, camera, microphone and all fittings.

What’s left is to design the motherboard to fit inside the plastic body.

The feeling of building your very own laptop by yourself is incredible.

Every one can go to the shop and buy laptop, but to build one with your hands, and to know every component inside is different experience.

We start seriously thinking to make Do-It-Yourself kit version where you get all components and instructions and can build your laptop.

Why not choose different boards with different SOCs and configurations which you put up to your choice.

These who come to the Help Day this Saturday will be able to touch for first kits, we have one in White and one in Black.

Crossing fingers to may complete the mother boards to show at FOSDEM in January 🙂

85 Comments (+add yours?)

  1. Bilal Tonga
    Nov 24, 2015 @ 19:32:15

    you are very quick! i do want a kit to build my laptop 🙂


  2. Marcus Moeller
    Nov 24, 2015 @ 19:45:41

    +1 what about a small crowdfounding campaign?


  3. KiLLWarez
    Nov 24, 2015 @ 20:23:09

    Great idea to fit boring devboard to laptop case.


  4. freedomcalyCaly
    Nov 24, 2015 @ 20:33:02

    Awesome! What I would like to see is a webcam and microphone that can be turned off by a proper switch that no software can manipulate, and a LED that shows when they are in use, that is equaly tamper-proof.


    • Lucio De Re
      Nov 25, 2015 @ 08:52:19

      Good catch, indeed!

      My thinking is to somehow approach this like an old-fashioned Hi-Fi kit: have a few options that can be chosen from for different purposes.

      An ARM, rather than i386, motherboard would be one idea, different modules for Wi-Fi, Bluetooth would be another. An external generic bus (Olimex’s own, even) might be very useful.

      I wonder if one could expect the chassis to outlast the motherboard, thus justifying a series of progressively more powerful devices in the same box?



  5. zuhku
    Nov 24, 2015 @ 20:53:06

    Good to see the progress. However, I’d still like to see alternative keyboard design with non-squeezed arrow keys. The squeezed arrow keys are unusable (and ugly to boot).


  6. anonima
    Nov 24, 2015 @ 22:15:51



  7. Mongrol
    Nov 25, 2015 @ 00:47:11

    How usable is it without Allwinner firmware blobs?


    • LinuxUser
      Nov 29, 2015 @ 06:23:47

      I guess it would be eventually usable without blobs, except 3D acceleration. I mean 2D would likely work, mainline kernel already supports “simple framebuffer” thing for most allwinners, and there is also KMS driver in works. Though since ARM are bitches, they do not release specs or sources on instruction set and internals of Mali GPU, so when it comes to shaders generation and somesuch, you’re out of luck, unless someone manages to reverst it and write driver. And somehow ARM GPU isn’t really inspiring and seems to be several generations below of modern designs. So I really do not get why ARM have to be such a btching company in regard of GPUs. Competitors can’t steal anything – nobody in sane mind would steal such a shitty GPU design in first place.

      Just to give you idea, more or less recent PowerVR GPUs are going to support brand new Vulkan API. Let’s see if ARM can do it as well :).


      • Mama Luigi
        Dec 02, 2015 @ 22:13:46

        When it comes to Vulkan… not on Mali 400. From what I heard anything with OpenGL ES 3.1 support will be able to use Vulkan, but the Mali 400 is a REALLY old, and according to Notebookcheck it was announced in 2008 and only supports OpenGL ES 2.0. Seriously, Allwinner is shipping a SoC with a 7 year old GPU design they’ve been using since 2011 (possibly earlier). According to their website, the graphics in even the old Allwinner A20 should be on par with this, unless the clock speed is significantly higher, but either way, still the bottom of the barrel.

      • LinuxUser
        Dec 13, 2015 @ 07:49:17

        > From what I heard anything with OpenGL ES 3.1 support will be able to use Vulkan
        I would not take it as granted. Basically, Vulkan is lower level interface compared to GL, it allows one to manage GPU resources without GL runtime getting in the way. Somehow, that’s what most 3D devs really want, especially those creating gaming engnies. Most GPGPU-capable GPUs with unified shaders are technically ok for Vulkan. But subject to willingness of someone to write driver.

        Mali is prehistoric thing, they do not have unified shaders. They have separate vertex shaders and fragment shaders and these are different. Such design would not do, it just ancient. PCs ditched non-unified designs like 10 years ago, but ARM still dares to offer their dinosaurs shit and even being bitchy about source or documentation. Amazing. So yeah, there is heck a lot of room for improvement and I guess ARM would eventually face what they deserve: competitors.

  8. 99guspuppet
    Nov 25, 2015 @ 00:47:46

    I want one


  9. 99guspuppet
    Nov 25, 2015 @ 00:51:54

    Is the keyboard replaceable ?

    what I want …
    == fast swappable battery “FastBatt” ideally attached to bottom of laptop ( I like the idea of a stackable laptop )
    == BlueTooth for adding peripherals
    == NFC
    == ports for adding cameras to the inside / outside of lid
    == ports for i2c, UART, GPIO on bottom of laptop ( 777stackableLaptop )
    == USB3.0

    ( 777DIYLaptop )


  10. SK
    Nov 25, 2015 @ 00:57:33

    How many mAh and at what voltage is the battery?


  11. Trackback: One Education’s Infinity is a Rugged Modular Laptop for Kids (Crowdfunding)
  12. liums
    Nov 25, 2015 @ 08:09:51

    can we hope 3d sources plan ( sorry i didn’t got a good english i don’t know how to formulate it ) for the plastic laptop case ? ( to then use a 3d printer )


    • OLIMEX Ltd
      Dec 01, 2015 @ 09:21:38

      sure, but I doubt the quality of the 3D printed laptop will be same as injection molded parts, some parts are very thin to become lighter like the inside frame of the LCD and I doubt it can be printed well by printer at all


  13. eduardo
    Nov 25, 2015 @ 10:23:50

    You might want to add a FPGA to your motherboard which is programmable by our opensource FPGA toolchain- the toolchain even runs on ARM.

    It would make a great hardware hacker laptop making available external FPGA connectors (like PMODs,… )

    you find infos about toolchain here http://www.clifford.at/icestorm/

    We work on a complete software support package to handle the FPGA hat from the RaspberryPi. https://www.youtube.com/watch?v=DHAthocTceg

    With our FPGA toolchain for the Lattice FPGA and our board you can do things like https://twitter.com/oe1cxw/status/660870171502997505


    our webpage is http://www.icoboard.org

    please lets talk

    we are at 32c3 in Hamburg


    • LinuxUser
      Nov 28, 2015 @ 18:59:42

      Whoa, finally someone reversed at least some FPGA. Would be great if you would support various hardware, not just Pi. Especially granted rather closed/proprietary nature of Pi.


      • eduardo
        Dec 01, 2015 @ 09:37:50

        “Every board” is supported by the FPGA. Its just that the icoBoard is connector compatible with the RasPi for ease of use and because there are so many RasPi. Take a connector cable and our FPGA board will work with any host-board.

  14. eduardo
    Nov 25, 2015 @ 10:25:41

    Here you can see how to use the FPGA toolchain.


  15. Bib
    Nov 26, 2015 @ 10:53:42

    Good to see that kind of things is possible and you did it. Great.


  16. davidak
    Nov 27, 2015 @ 20:47:06

    a open hardware laptop would be amazing :>


  17. LinuxUser
    Nov 28, 2015 @ 14:17:17

    Nice idea! Though to make it a decent desktop, one may want SATA drive, to attach fast SSD.


  18. 99guspuppet
    Nov 29, 2015 @ 00:54:21

    RaspberryPi PiZero sells for $5 USD.


    • LinuxUser
      Dec 13, 2015 @ 16:23:14

      > RaspberryPi PiZero sells for $5 USD.
      It serves as example of fraudulent marketing. Because you have to pay some extra for S&H. And good luck to buy it without cables, worth of $20 (hmm, isn’t it too much for just some wires?). So they can declare even $0 price, but good luck to actually buy it at this price.

      So you going to be damn lucky if total amount of money you ACTUALLY pay for PiZero crap is below $25. That’s how you do marketing bullshit. Not to mention PiZero hardware is utter crap. So it only makes sense if you can buy it at $5 in bulk, and good luck to do it, dammit. Buying single core ARMv6 (ARM11) crap at $25 is just a ripoff. China boards would have quad core and gig of RAM at this price, with FREE S&H. So you can get much more for same $25 closing the deal.

      At the end of day it seems Broadcom is trying to get rid of stock of their outdated crappy ICs while trying to make some margins. As usually: good marketing, crappy engineering. But its nice it would put some extra pressure on chinese manufacturers, who would have to reduce prices. And, erm, Olimex, you’re warmly welcome to do some cheaper things, too 🙂


  19. 99guspuppet
    Nov 29, 2015 @ 01:06:24

    I am looking at https://olimex.files.wordpress.com/2015/11/laptop-kit.jpg

    Would you publish more photos of the laptop case ? Closeups of the front and back pieces ? ( For example: a photo of the backside of the keyboard ) What is in the bags marked “camera” ? Is the battery fixed in place ? Are those speakers at the right and left of the base ?


  20. 99guspuppet
    Nov 29, 2015 @ 01:45:34

    Maybe a USB 3.0 connection through the base. The connection would be made to with an add-on board with spring loaded contacts. Board would be secured magnetically. Also power can supplied to the laptop or borrowed from the laptop using a couple more spring loaded contacts.


    USB 3.0 is ubiquitous ( okay , maybe USB 2.0 is more ubiquitous )


  21. 99guspuppet
    Nov 29, 2015 @ 05:28:26

    I put a better link to my drawing…. in .pdf format https://drive.google.com/file/d/0B9IHkDDdUcjQaG5EbElpeU1KWXM/view?usp=sharing


  22. ignat99
    Nov 30, 2015 @ 13:45:34


    It is a local Russian Oberon community. I think they could use this device to their alternative projects.


  23. LunuxUser
    Nov 30, 2015 @ 18:13:01

    Btw, Olimes, you’re literally reading my mind: I’ve got fed up with Intel ME firmware backdoors, etc and got to think my next laptop is not going to be x86 at all. Now it seems to be more real than ever. Woot!

    Though I’ve seen it like this: grab CPU module, design low-tech 2-layer daughterboard to place sockets, etc matching enclosure’s geometry, solder module on it (like it happens in some chinese tablets). Obvious advantage is that I have to design low-tech 2-layer adapter board, which is far less challenging than doing high-speed multy-layer board and also cheaper to manufacture. The obvious disadvantage… uhm, well, opensource CPU modules are yet to be seen, so I couldn’t claim such design is all-open :(.


    • OLIMEX Ltd
      Dec 01, 2015 @ 08:29:13

      USB 3, HDMI etc differential signals is near impossible to be routed on 2 layer board as you can’t make controlled impedance differential routing in reasonable way there.


      • LinuxUsers
        Dec 13, 2015 @ 04:50:27

        Hmm, that’s what I like about your blog: really good way to learn some unobvious things.

        Though from practical standpoint…
        1) USB3 isn’t exactly widespread to the date. So I haven’t even planned for it so far. Are therer any cheap, fancy and open CPU modules with usb3 around so far?
        2) Not like if I really need HDMI out as well, because laptop-like device is supposed to have its own screen anyway. And it clearly lacks horsepower to handle worthy displays anyway since they are demanding in term of “horsepower”. Though it good to have.
        3) Granted both usb3 and hdmi are working via cables and these are hardly perfectly matched, since users are doing all sorts of nasty things like bending, etc – I guess there is quite some tolerance. And I guess careful 2-layer design can be okay, no? Furthermore, multi-layer board on its own is far more complicated and I guess it actually haves more reasons to expose non-uniform properties, thanks to prepregs and somesuch. Since module would do all hard parts, low-tech part can be so simple I can, say, afford ground plane on bottom side of lanes and differential lanes on top part of that. After all it just have to connect module to output and there could be plenty of space, so I guess one can get more or less adequate lanes, no? At least it works for Ethernet. And usb2 proven to be so tolerant it has worked even over short piece of Ethernet twisted pair, which IIRC comes with different impedance and no shielding, which is mandatory requirement in USB2. Its really wrong thing to do, but it worked and I haven’t even detected bus errors. Though it’s clearly does not serves as good example of engineering and only happened as “last resort” thing.

  24. Shiba
    Nov 30, 2015 @ 23:21:05

    Is Allwinner mandatory?


  25. John
    Nov 30, 2015 @ 23:45:55

    an e-ink display option would be killer! I’ve always wanted a laptop with an e-ink display for working outside (long battery-life, sunlight readable, etc).


  26. Luis
    Dec 01, 2015 @ 00:10:54

    Hey Olimes! I can’t wait for you to finish it. The world needs something that does not spy or betray their users or otherwise just turns into a brick a fews years later because there are no drivers nor documentation…

    Could you start something that would allow others to build it too? Maybe go like the open pandora projetct? Or maybe lowrisc.org project? Maybe just crowdfunding?

    Maybe Chinese manufacturers will realize that they would sell zillions of gadgets if only the community could feel at easy to tinker and improve the software (all being free and open source, of course).

    Anyway, Great idea.


  27. djmdjmdjm
    Dec 01, 2015 @ 00:34:19

    While everyone is dropping wishlist items, I’ll add mine: a OEM ThinkPad keyboard

    (oh and a 4:3 screen)


  28. Danny
    Dec 01, 2015 @ 00:50:52

    Nice to see the Linux Laptop idea taking form 🙂

    I have just one request: if possible, please don’t use RTL8723BS.

    I think that if this is supposed to be an OSHW laptop there shouldn’t be closed-source firmware required on it.

    Because when I use a closed-source laptop I’m careful what I do on it and always mindful that everyone can see now or much later what I did on it. But on an OSHW laptop I’m more trusting.

    Having a black-box updateable firmware is thus potentially much worse on the OSHW laptop, and on the LAN communication pathway it’s even worse.

    I would suggest one of the Atheros chipsets which have GPL firmware (by Atheros themselves) or another chipset which has non-updateable firmware (I mean in the physically-impossible way) or GPL firmware or no firmware.

    With the world being as bad as it is lately, please don’t make it easy to spy on us. Sigh, what a time to live in where that is not a conspiracy theory 😦


  29. Bill
    Dec 01, 2015 @ 04:57:13

    So not interested in a blob-laden ARM laptop. Say goodbye to ever upgrading the OS.


    • OLIMEX Ltd
      Dec 01, 2015 @ 08:24:26

      What you say here is simple not true. Allwinner devices boot first from USB and SD-card so never could be bricked and always can be re-flashed with new OS without any trouble.
      And this blob mania becomes a bit boring for me, no one ARM device is blob free, why Allwinner’s one should be? If you want to use blob free software you can do it right now too for most of Allwinner devices, just do not include few drivers in the kernel.


      • Danny
        Dec 04, 2015 @ 21:20:25

        I think Bill meant binary-only blobs in the Linux kernel (if those exist for A64), I agree with Bill on that that would break on Linux updates.

        To expand on what OLIMEX Ltd said to Bill:

        on Allwinner A10 and A20, from the boot0 (its source is available) first-stage bootloader running in ROM all the way up to Linux userspace, all the things are either fixed function circuit or Free Software, except Mali 3D (and there only the Mali userspace library is questionable). That includes SATA, 2D graphics, audio, video decoding, TV in and out, CAN bus, IiC, SPI, crypto etcetc. There’s documentation about pretty much everything.

        (With Allwinner, it was much worse in the past, but nowadays it’s pretty good – for the A20 model)

        It’s also unbrickable since it always boots from microsd if one is there. Also, there’s FEL in order to transmit whatever kernel you want via USB (then can continue via NFS) on power-on.

        That’s the main reasons I’m using them. I really don’t want to wake up one day to my appliances (which in some cases I built) turning against me – like in a bad science fiction movie turned real. If i then reformatted the appliance and it was still going against me, that would suck (and happen).

        How it is with A64 still remains to be seen – but Allwinner are a Linaro member now so maybe someone let them in on why, among other things, this whole Free Software thing is being done. (I’m not holding my breath, though)

        As for me, it depends on where the blob is. For GPS connected via serial, a firmware blob is OK for manual operation since GPS is receive-only and all an adversary could do is make it tell me the wrong location. But they also could do that by just modifying the EM waves that reach the device – and that doesn’t even require any access to the device. And since I use it manually I won’t believe stupid locations it tells me – so what’s the use for the adversary? So basically the risk doesn’t increase.

        On the other hand, having an updateable firmware blob for (say) Ethernet pretty much means I have to https *everything* (if only for the passwords). I would probably forget some things and so the risk would massively increase and something like that is an instant no-go. (in the future probably everything will be https or ipsec and so it will not matter anymore whether the link layer is a blob or not – but today is not that day)

        That said, it took me a long time to reach this point. I can see how it can be annoying to have to take care about these largely non-technical matters (let’s face it, these attacks are pretty much a social/political problem). Hell, *I’m* annoyed I have to deal with that crap.

        But that’s the world of today. Try connecting a machine to the public internet and wait until the first cracking attempt is made (by nations, presumably). I measured 30 s. If they broke into the machine, I’d like to minimize the places they can hide in.

        >And this blob mania becomes a bit boring for me, no one ARM device is blob free, why Allwinner’s one should be?

        Because it is possible to use A20 without firmware- and kernel-blobs right now. By being blob-free they attracted customers that cared about that (basically anyone paying attention to the post-9/11 world and lots of Free-Software people). Now the upgrade path of these customers is of course also required to be blob-free.

        But enough on that. I’m first and foremost a hacker in the old sense of the word – so, please, let’s have fun and hack on. Sorry for having been the annoying killjoy in this case 😦

      • Francisco Franco
        Dec 07, 2015 @ 11:15:58

        “As for me, it depends on where the blob is. For GPS connected via serial, a firmware blob is OK for manual operation since GPS is receive-only and all an adversary could do is make it tell me the wrong location.”

        This makes no sense to me. GPSd handles the $GPS stream.. No blobs needed. All NMEA0183 output is a stream of Lat/Lon coords plus some.
        I do agree blobs stink and Atheros Wifi is king of openness. Also kudos to Sierra Wireless on the Cellular modem end, with no blobs.

      • LinuxUsers
        Dec 13, 2015 @ 05:24:55

        Actually, I have a blob-free boot sequence on Allwinner A10/A20 things and I consider it a MAJOR advantage. Though I do not care about GPU in most cases, my things are usually automations and somesuch, so I rather care about security and predictable attitude :). It is really not going to be cool if someone would be sabotage some automation device in unexpected ways using backdoors, etc.

        The only “blob” I have is on-chip boot ROM. This one is simple and well-understood (linux-sunxi ppl almost completely disassembled it). Then it kicks my (opensource) u-boot. This one in turn kicks my (opensource) kernel. This kicks rest of system. Everything is open and under my complete control. Which is great and so I can undertake personal responsibility even in some quite demanding use cases, being ready to provide complete source code of virtually everything to prove there are no backdoors, etc. And I can change everything, being able to target strange tasks. Say, in one case I want to use UART to communicate with module and it is really bad if someone would write some unwanted strings to UART. In fully open stack I can chop any UART output. And maybe add some non-default reaction on e.g. button press, etc. If it going to be blob, is just not going to work right and I would eventualy hit a brick wall with my face, when I need device to do something, blob can’t do it and I’m not allowed to change blob either. This isn’t good thing for sure. Linux happens to be what it is thanks to openness. Good luck to customize Windows though. That’s what you get with blobs at the end of day.

        And btw, ROM boots like this:
        1) One of SD-cards is top priority.
        2) NAND.
        3) Other SD cards, SPI-NOR, etc.
        4) If boot code is missing in all storages (empty system) or u-boot button pressed, simple USB stack starts and can be used to inject initial loader to RAM. This is convenient for initial flashing and somesuch. E.g. tablet should boot from NAND and after hardware is assembled, one have to get software to NAND somehow. Sure, one can program NAND before soldering. Or just plug cable, power it up and flash in place, which is easier and less demanding. One can also do various tests, e.g. memtest this way. So this idea is quite widespread.

        As for blobs… blobs were abused a lot to place anti-user features and just backdoors. E.g. intel ME firmware is quite nasty. Supermicro gone further and has placed blatant ME backdoor aka “engineering login” and even has been caught on it, so Supermicro server owners can expect unexpected visitors if they dare to connect ME-capable port to Internet, everyone can reinstall or pwn OS, etc. Virtually all home routers got the same thnigs in stock firmwares. Are you getting idea why we are not fond of blobs and welcoming community projects which are truly open?

        Right now I’m looking on root prompt on some router. No, I do not know user name and password, I do not even know where it located. I’ve just used “engineering” login, because I’ve got curious. And so, if I would decide to be a bitch, I can, say, dd /dev/zero to /dev/mtd, trashing boot block and leaving it a perma-brick, which can only be recovered via JTAG (which isn’t what average mortal can do). All this … over the network?! Of course I can also intercept/hijack all traffic, change settings, etc if desired. Though since I’m not a blackhat, I guess I’m better to alert device owner their router suxx. Whatever, I do not think it should work THIS way. So let’s deblob the world.

  30. Potrebitel
    Dec 01, 2015 @ 07:33:48

    Congrats Olimex,

    you made it to top page of Hacker News !

    For me thats a good achievement, so keep going in the right direction.


  31. Christian Nobel
    Dec 01, 2015 @ 13:43:29

    Sounds interesting.

    What about keyboards in other languages than English?


    • aleksphp771
      Dec 14, 2015 @ 13:07:38

      For Cyrillic you can use keyboard stickers, search around on sites like eBay or professional sellers you might find them for your language. English is either way a must have on a keyboard.


  32. Francis Kim
    Dec 01, 2015 @ 14:04:51

    Looks great 😉


  33. iiiears
    Dec 01, 2015 @ 15:44:36

    Looks great,


  34. JoBBo
    Dec 01, 2015 @ 20:17:13

    Hello, what type of display is being used? Is it a TN panel? Also, what is the resolution and size?


  35. Svetlio D (IGNITER)
    Dec 06, 2015 @ 23:25:04

    There is something in the air, the soil or the water in Mayna Town that makes the people there genius 🙂


  36. Rui F Ribeiro
    Dec 08, 2015 @ 14:47:40

    When will you make the notebook available to the general public?


  37. Ne3s23p2
    Dec 09, 2015 @ 22:51:12

    What about ergonomic keyboard like TypeMatrix ? Or at least an alphabetic keyboard with square grid layout, not staggered ? Once for all, don’t allow resistance to change, and ditch ancient qwerty layout (a show stopper for me).
    As some want to avoid a builtin camera and microphone, an HDMI video input should be an alternative, with embedded or separate digital audio I/O. Wireless should be optional too.
    Hardware hackers will like lots of GPIO including SPI, I2C, UART, a few analog inputs (ADC, RTD & TC sensors…). Network hackers will like at least 2 ports, maybe with SFP slot.


    • Lucio De Re
      Dec 10, 2015 @ 07:02:19

      This hacker would like USB to “go away”!


      • 99guspuppet
        Dec 13, 2015 @ 00:02:50

        LDR What does your world look like without USB ? Gus in Denver

      • Lucio De Re
        Dec 15, 2015 @ 06:44:08

        In short: PoE only. Of course, all these neatly incompatible devices will become inaccessible, but I should not be the only one wishing them away.

      • 99guspuppet
        Dec 16, 2015 @ 03:53:45

        Are you suggesting POE at 100GbE ? Sounds good …………..

      • Lucio De Re
        Dec 16, 2015 @ 05:47:03

        I’m just telling you what I wish for. I’d be rich if I could implement all the hare-brained ideas I have come up with in my life!

        That said, if we standardised on ethernet (or even token-passing) as the networking standard, with PoE where power needs to be centralised and Wi-Fi (you’ll need to look up the relevant IEEE codes yourself – 803.11 and friends or equivalent), we wouldn’t need puny Bluetooth and poorly defined USB categories at all. You sound like somebody who understands where I’m coming from.

      • 99guspuppet
        Dec 16, 2015 @ 06:02:56

        I can see merit in your idea…. It would be fun to see what happened if some big hitters ( Intel , Apple, etc. ) adopted it.

      • Lucio De Re
        Dec 16, 2015 @ 09:05:31

        That was a choice BEFORE firewire, USB or Bluetooth. Today, it would need adoption by a different, less commercial community with a lot of clout.

  38. Stupido
    Dec 10, 2015 @ 13:49:59


    • LinuxUsers
      Dec 13, 2015 @ 05:30:30

      I guess this one isn’t going to be open hardware. Furtheromre, if you getting stuck with hardware-related stuff, only really few humans on this planet can give you a proper answer on how they created their hardware, how it works and why it happens this way. Olimex is one of such a bold humans around. That’s what makes it really pleasant to use their boards. This is even worth of some small extra, since Olimex expertise clearly counts as “added value”. Though cheap units are good and have more uses. So cheap prices are warmly welcome. But preferrably from Olimex :P.


  39. Trackback: Allwinner A64 Android 5.1 SDK and Linux Source Code
  40. 99guspuppet
    Dec 13, 2015 @ 00:04:05

    If this will be OSHW…. may we see the preliminary data ? the current design and datasheets ? I am very excited ……


  41. aleksphp771
    Dec 14, 2015 @ 13:05:30

    Hello, I was wondering a lot about the Allwinner SoC and there is pretty much no better place to ask than at OLIMEX. This laptop is very interesting, it runs the A64 right? What is the difference between R18 and A64? Is it the same thing maybe? I think. These Allwinner chips are great value for its price but the problem is they were made to run Android (like I heard people here have complained about before) and not mainline Linux. With projects like C.H.I.P etc increasing the popularity I wonder if any Allwinner chips supports running mainline Linux without any horrible kernel patches, like this A64? I am very interested in being able to power the ARM with Debian or Arch Linux but Allwinner never intended people to do mainline on any of these. This, at least, sounds extremely interesting. If this laptop can run Debian or Arch Linux I would definitely love this kit. What OS will it support though? A normal laptop with an Intel CPU can support pretty much all kinds of OS but this Allwinner probably cannot. Still, the choice to build your own laptop, hmm…


  42. Thomas
    Dec 14, 2015 @ 20:46:04

    You’re free to do with Olimex’ laptop what you want. At the moment you’ve to rely on kernel 3.10.65 (and can combine this with your ARMv7/ARMv8 rootfs of choice) but you could also use ‘Remix OS’ on it or stock Android.

    What I’ve seen so far in .dts and sys_config.fex files it seems the A64 is prone to overheating. And while Amlogic’s S905 looks like the nicer hardware (faster, maybe less thermal issues) I doubt we’ll ever see a more recent kernel there (they’re at kernel 3.14.29/u-boot 2015.01 — maybe forever) but I would suspect we’re able to boot mainline u-boot/kernel next year on the A64.

    There’s no need to like Allwinner but the linux-sunxi devs do a great job again and again (and that’s my main reason to choose Allwinner hardware in the first place)


    • aleksphp771
      Dec 14, 2015 @ 22:40:27

      I guess you are right, but the laptop would probably benefit from GPU acceleration if it could have gotten that. Remembering the statement about OLIMEX picking Allwinner because of the linux-sunxi devs too in some other post. The main problem is that I have bad experience with Allwinner, the CPU overheats and is not able to perform its theoretical speed – the specifications that are on paper which is 4 core 1.2GHz, which is something I’d really like to use while being in a power saving environment. Not only does it save power costs but it can extend battery time greatly. Is it not clocked a little too high again? If I am able to use this clock speed at max for some around a hour at random times I would definitely pick it as my use is centered around CPU usage. Also, I am unsure about R18 and A64, are these the same? Why are they both labeled at A64 then?


  43. Thomas
    Dec 14, 2015 @ 23:29:03

    Regarding the name scheme: no idea, R18 seems to be similar to A64 to some extent (the Pine64 guys started with R18 and switched then to A64). Same with A83T that is now known also as H8 (with different PMU) and R58.

    All modern SoCs implement thermal throttling or ‘budget cooling’ and that works quite ok if you don’t overvolt the chips (like the Orange Pi maker did to advertise the 1.2 GHz H3 as “up to 1.6 GHz” SoC). I digged a bit deeper recently with A83T and H3 and if settings are ok even the H3 stays cool (again: http://forum.armbian.com/index.php/forum/10-free/)


    • aleksphp771
      Dec 15, 2015 @ 10:02:32

      Looked around a bit too. If I can just heatsink the A64 and it won’t overheat with normal clock speeds I would be happy to use it. I overclock my Pi and use three heatsinks on it (overboard) and that works out pretty well. I also have some thermal paste I can use but those are mostly for bigger CPUs right? What I’ve read confirms what I thought, that these are made for less consumption and higher clock speeds which is why they might act like older Cortex SoCs, I don’t know much about the process behind these, wafers and that, but I’m sure there is a reason it was designed this way. If it does give me the 4-core 1.2GHz, which it might with the correct throttle settings and heatsink, I will definitely give this CPU a go. It would be very interesting if anyone can provide any tests of how this CPU performs but that would require somebody with the CPU to join in and perform the right tests.

      A prime example and probably why I am a little doubting is that Snapdragon didn’t exactly do well when they came out with their 64-bit series, the news started rolling in with the CPU overheating and getting heavy throttled until they upgraded the cooling system and the manufacturers pushed updates that throttled the clock speeds even more. As my current CPU OC to 1 GHz is only single core this quad-core would not only run on the same frequency on regular clock but it will also have more cores and be 64-bit, which is going to be a great advantage in speed, so I am considering buying an Allwinner despite their outdated kernel. If they give me mainline I would probably have been sold to give it a try already.


  44. Thomas
    Dec 15, 2015 @ 12:25:51

    It seems I might get my hands on a ‘Pine64’ soon so when I find some time to test I’ll report back.


  45. Khalil Gibran
    Dec 20, 2015 @ 15:40:33

    Reblogged this on kgibran and commented:
    Hopefully this will support FreeBSD…


  46. Dimiter
    Feb 28, 2016 @ 17:02:46

    The first bulgarian laptop was build in the 88 in Plovdiv and called “Pyldin”
    Why not call your design Pyldin-IIe?


  47. Will Hart
    Mar 15, 2016 @ 02:00:36

    Truly cool man I am waiting for a pine a64 +2 board myself can I ask you a simple question sources for a display for one. Thanks


  48. notzed
    Mar 28, 2016 @ 12:31:54

    matt screens? 4:3 would be nice too.


  49. thomas
    Mar 31, 2016 @ 19:28:52

    This laptop is a great thing 🙂
    Do you have any idea when it might becomes ready for production ?


  50. khertan
    Apr 21, 2016 @ 15:34:02

    Any news ?


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: