Software for the Raspberry Pi is quickly moving forward. Beyond the several core Linux distros, another couple dozen systems are available, with NetBSD, FreeBSD, and Chromium imminently stepping into the mix. (Ubuntu will not join them as it requires ARMv7 and the Pi is ARMv6). Two dozen programming languages are available, including Python, Perl, Java, Ruby 1.9.2, BASIC, and more. Since the Pi
is a full fledged ARM computer, it should run nearly any ARM app within its system requirements. See the RPi Wiki or Foundation website for more info.
I’ve read recently on the RPi website that the creators give the project a three year shelf life before it ends as we know it. Whether that means a new, more advanced device at the same price point or the actual end of the experiment altogether is what I wonder about.
I haven’t picked up one of the Rev. 2 boards yet, as so far I haven’t had any issues with my “original” board, but I do plan to get one soon just for the doubled RAM. That aspect alone has made a difference in a lot of projects documented on the forum. I’m hoping for continued improvements to the USB circuitry in a third revision if one should come to pass.
One thing is for sure: The release of the Pi has spurred competition in its price range. There are no less than four excellent ARM boards for under $60 out there now, with varying degrees of F/OSS support via open drivers. If Broadcom will ever open up the drivers for the BCM2385* we could have perhaps the most versatile SoC board in existence. I’m not holding my breath, but I’m not giving up on the platform either.
*Unlikely as long as commercial devices like the Roku 2 continue to use it in a proprietary fashion.
Don’t forget the horde of Allwinner A10 based boards, phones, and tablets! The A10 is pretty much entirely open now that people have Free drivers for its GPU core.
Well, I’ve always understood the A10 to be the go-to chip for cheap Chinese knock-off tablets. However, it seems to hold its own especially when you consider the price/performance matrix. Given the openness too, I’d love to get my hands on a Pi-like dev board based on it.
Edit: Well that didn’t take long…
http://www.cnx-software.com/2012/08/31/49-cubieboard-allwinner-a10-…
Edited 2012-11-24 06:59 UTC
Why do you want a board based on the A10? The GPU is completely locked down, there are no open-source drivers for it, making it quite a risky choice. Limadriver is not a solution as it’s still years away from even becoming stable enough for 2D desktop operation.
Edited 2012-11-24 10:07 UTC
I thought the Lima driver was further along than that. What would you recommend that might be in the price range of the board I linked? I can’t afford a PandaBoard.
At the moment I cannot recommend anything. They all come with closed GPU, meaning that whenever the vendor gets tired of updating the drivers you’ll be stuck.
So how does this closed GPU situation impact you with the boards and devices you play with?
I suppose I should stick with the Raspberry Pi for ARM fun then, and possibly look into the Loongson based solutions for the long term. Either way, I’m slowly weaning myself off of x86. I suppose I’ll hold onto my “classic” rig for BeOS/Haiku/Win98 gaming purposes but my main computing devices will one day be devoid of x86 and Windows.
There was a few weeks back a mini-story here about three different ARM-based boards, similar to the RPi, and I wrote there in the comments about my experiences with my Pandaboard. See http://www.osnews.com/permalink?542309 Or if you don’t wish to read the whole comment I’ll just say that the current situation sucks arse.
I really like all the possibilities and the general idea that these small, cheap boards represent, but until there is a company out there that is actually willing to maintain a properly working set of drivers for their chips and/or willing to release fully-working F/OSS drivers I just cannot quite recommend anything. I wish the companies would see the potential and commit themselves to enabling us — as a community — to take the scene where it belongs.
Edited 2012-11-25 06:26 UTC
I don’t know if you wish to continue the discussion any longer, but well, the Cubieboard ( http://linux-sunxi.org/Cubieboard ) is a great board hardware-wise: it’s got powerful enough CPU and GPU for most desktop-tasks, it’s got a SATA-connector for using real HDDs, there’s not any particular component on it that gimps the rest, and the expansion headers provided on the board include all the most important features — I particularly like the inclusion of SPD/IF – header there. The board is also cheap at only $49, meaning that you wouldn’t break any budget even if you bought everyone in the family a box of their own.
Alas, the board is also a great example of how badly the software side of things can ruin an otherwise terrific product, what with things like the Mali-400 – driver sometimes actually being SLOWER than even a FBDEV at basic 2D operations, the drivers either missing X11 – support or OpenVG and/or OpenGL ES being broken, and so on. (See http://linux-sunxi.org/Mali400 more on that!) Similarly, H/W video acceleration is downright useless in that that only applications that directly utilize the CedarX – library can use it; no Gstreamer-support, no OpenMAX, no nothing. An example of what no standard method for accessing the acceleration – features means is VLC: VLC only works from console, no GUI, and since Cedar lacks YUV420 there’s no OSD, either! ( http://linux-sunxi.org/VLC and http://linux-sunxi.org/CedarX )
The above is exactly why I actually try to recommend people to stay away from these things for now; sure, all the hardware-features look great on paper and it’s easy to get excited about these, but once you actually try to use the thing you’ll quickly write it off as a failed investment on your part!
Thank you very much for that detailed analysis! I was afraid such a deal was too good to be true. I think I’m going to spend that $50 on the rev. 2 RPi instead, and donate my old one to a friend who wants to get started with programming and sensors/motor control.
Looking ahead, I’m going to be selling some of my superfluous gear (laptops and such) and look into getting a Loongson-based mini PC. It’s a platform I haven’t had a chance to work with yet, and it promises fully open hardware and software. We’ll see if Stallman is right about it.
That sounds like a worthy cause. One can never learn too much!
Did I mention I accept donations…? *cough*
If you ever get your hands on one how about writing an article on it here on OSAlert? I’d do that myself, but it seems unlikely I’ll be getting one all due to shortage of cash!
Yep, have 3 raspi’s sitting in the cupboard collecting dust. I replaced them with mini-itx boards with i3 low power processors. These use a bit more power at 35 watts, and cost a little more (but not much) but at least they do what they claim and work ten thousands times better and I can run whatever I want on them.
The whole arm thing on linux for general use is a scam!
Any suggestions you could link to for inexpensive stuff that’ll run XBMC and some emulators well?
(At least Snes9x and preferrably Mupen64Plus since our N64 controllers wore out but PCSX2 would be nice too since we’ve worn out Sony’s patience for free PS2 DVD drive replacements)
My brothers really want a home theatre setup but, once you factor in the heat from an old big-screen HDTV and an old 5.1 receiver, the high-TDP Athlon64 3300+ that they got for free makes it too much of a heat source to use in the summer without air-conditioning.
Switching to a really cool-running HTPC would probably be enough to solve the problem.
Edited 2012-11-26 09:31 UTC
I am not sure how the current situation in the ARM graphics space differs from say Nvidia’s binary blob on x86. Not that I am advocating closed source drivers.
There is some speculation over on Phoronix about Imagination Technologies open sourcing their PowerVR graphics in the near future. I hope this comes as a neat surprise to you.
Well, for one, NVIDIA is actually quite committed to keeping their drivers working and they expend a lot of effort in fixing bugs in a timely manner. On the ARM side of things the manufacturers do a few crappy releases and don’t really care if a release breaks something that worked before. Just go and take a look at the link I posted earlier about the state of the Mali-400 driver.
I would prefer F/OSS – drivers, but I have no qualms about using closed ones, either, I just wish the companies actually cared about the quality of their drivers! Oh, and kept the drivers alive for more than 6 months.
The thing is, closed-source drivers that don’t break is somewhat alien to the way Linux kernel development works. After all, who needs such a things as a stable ABI when you can just ask all low-level devs to bundle everything and the kitchen sink in the huge kernel codebase instead?
I think that’s one of the ideas behind the Videocore IV driver design. Run everything proprietary on a co-processor and it’s not too hard to convince management to open-source all the RPC stubs that run in the Linux kernel.
You might well be right! Thinking of it some more, that design, while it might feel exceedingly convoluted from an engineering point of view*, is quite a clever way to hijack the Linux kernel’s policy
* If I’m not misunderstood, when using the modern Linux graphics stack, a graphics operation has to successively get through all these layers of abstraction : User-space code -> X server or Wayland -> OpenGL library -> Gallium backend -> Kernel driver -> GPU binary blob -> Actual GPU.
Edited 2012-11-28 07:00 UTC
Of course, that’s also probably one reason so many people take such offence at it.
It really is a likely avenue for a slippery slope toward side-stepping the spirit of copyleft licenses.
Free as in Freedom? It is worth considering in that case.
No, the person you’re replying to is misinformed. The reverse-engineered Limadriver is still far from useable.
thanks for the info. I will not buy then.
Not sure if I understand this about the drivers… I had gotten the impression that they’ve already made the drivers open-source. The part that’s still proprietary is the chip’s internal firmware, which is something you don’t need to access its functions.
I suppose it means you can’t fix Broadcom’s bugs for them, and you can’t repurpose the GPU for other computing tasks (OpenCL?), but those seem more like quibbles than real obstacles, to me.
The argument is that, if a firmware is so complex that there’s a GLSL compiler in it, then it’s a co-processor with a closed-source OS of its own, not mere firmware on a subordinate processor.
Heck, on the Pi, the GPU is responsible for loading the image off the SD card and setting up initial state before handing off to the CPU so, if you’re of the tinfoil hat persuasion, you could plausibly rant about the possibility of backdoors and the risk of techniques similar to Ken Thompson’s hypothetical compromised C compiler.
http://cm.bell-labs.com/who/ken/trust.html
Edited 2012-11-24 13:11 UTC
As further clarification. The GPU runs what you’d consider a graphics driver on an RTOS (threadX). The application processor running linux then sends commands to the GPU much like an application would send commands to the OGL driver.
Things that require direct access to the hardware like OpenCL, or even X acceleration (Render) cannot be done (without broadcom offering updated firmware).
Objectively, it is quite an interesting encapsulation, but ultimately a useless one if you want access to the hardware. Ultimately, the Rpi foundation was mischaracterising the open nature of their driver.
Edited 2012-11-24 13:25 UTC
But it *is* a very useful one if you want the ability to use the GPU from something that isn’t Linux+Xorg.
Aye, If you’re just planning on using OGL from an application it should do fine, just like any other blob. Just don’t advertise it as the most open graphics stack.
As far as an SoC with an actually working graphics stack goes, I think it technically is the most open, but only by a hair, and they all suck as far as openness goes.
(Yes, the Mali-200/400 has that open source project, but it doesn’t actually work.)
It’s just as closed as everyone else’s, but because the binary blob is elsewhere, everything running on the ARM side is open. That, at a functional level, makes it slightly more open (in the “open systems’ sense, definitely, but not so much in the “open source” sense). As I mentioned, ANY OS can use this, and it also means that if things in Linux+Xorg (or, if it becomes Linux+Wayland) change, the ARM side driver can be updated. This is more open than everyone else, where things may end up tied to specific versions of the Linux kernel and Xorg.
Edited 2012-11-24 21:54 UTC
http://www.red-lang.org
As a Raspberry Pi owner, I’m increasingly unable to view the device as anything more than a $35 promotional toy advertising Broadcom’s Videocore IV. You don’t get to really play with that chip — which by the way, drives the entire board and is the most interesting thing soldered onto it. The ARM gets initialized by the Videocore, the ARM merely exists as the interface users have to interacting with this black box.
You don’t like the fact the firmware blobs are buggy and you trivially lock up the board while developing? Is low-level access really the only way you’re going to develop applications interesting to you? Sucks to be you!
According to Liz Upton, one of the PR creeps from the Raspberry Pi Foundation has the following to say to actual developers complaining:
‘… keep trying to rustle up some outrage if it gives you a kick; I^aEURTMd recommend finding something else to do soon, though. We don^aEURTMt want you developing an ulcer.’
When asked: ‘Am I allowed to be outraged by the fact that it^aEURTMs not really open, since I see no mention of actual documentation for the hardware? Or by this: http://www.raspberrypi.com/mpeg-2-license-key/ ? Can I get outraged by that?’
She replies: ‘Well you *could*, but you wouldn^aEURTMt half look silly.’
(sourced from: http://www.raspberrypi.org/archives/2221 )
What a charmer. This is someone totally detached from reality and is so full of herself and her own product she actually thinks she can talk to developers that way. She has no awareness that nothing the RPi has done or offered is very groundbreaking, ARM SoC eval boards like the RPi were inevitable — the evolution of products like the Sheevaplug and other plug-computers in addition to hobbyist hits like the Arduino paved way for this.
Hardware is being eaten by software. Broadcom wants to produce an artificial value to their hardware by locking it up with their own crappy software and they want you to cheer for what an innovation this is. End of story.
Edited 2012-11-24 17:43 UTC
Interesting comment. I haven’t gotten into the Pi heavily enough to figure out all you have about the Broadcom chip. I wonder if the small-board competitors have the same issue? I’m gonna look them up and see if they all use Broadcom. Thanks.
There simply isn’t anything in ARM SoC space where this is not the situation at the moment. The closest we have are people attempting to reverse-engineer ARM’s Mali line of GPUs (used on that ODROID-X, for example) and the Lima driver currently is quite primitive, as WereCatf mentioned.
This is pretty serious stuff, though. Every day I’m more convinced that the war against general purpose computing is going to be won by merely cutting off access to proper video card support. How relevant can you really be to the average person if you can’t even handle a compositing desktop with the bells and whistles like they’ve come to expect? Oh boy, we have framebuffer access — I’m sure my family is going to love rocking Openbox like it’s 1999.
The fact is — right now there’s a $120 board that’s more than capable of handling most people’s desktop needs, except we can’t write proper software to make that fully realized. But feel free to gaze upon http://www.hardkernel.com/renewal_2011/products/prdt_info.php?g_cod… and wonder what could be.
Edited 2012-11-24 20:53 UTC
With only USB2.0 ports and no SATA all that speed in the CPU is pretty gimped.
The eMMC port is still better than SDHC and whining about USB 2.0 as if that’s make/break between my mother or other average people using it as a general purpose browser/email machine on a budget is hilarious. Thanks for the non-comment and your lack of vision.
Edited 2012-11-24 21:05 UTC
I wasn’t talking about your mother, now was I? The fact is that USB 2.0 HDDs are terribly slow, limited to about 35MB/s speeds, and as such boards like this still wouldn’t work for quite many people. Also, if we were talking about your mother she wouldn’t need a quad-core CPU or GPU for mail and browsing anyways.
No, but you also weren’t talking about the average user, either. Instead, you came to nitpick and bizarrely quote data transfer rates. That’s what you came to discuss instead of commercial applications of cheap hardware. Good for you, have a star.
What the hell? Why are you so terribly defensive the moment I mention possible bottle-necks? With a SATA-port the CPU wouldn’t have to spend so much time idling when trying to read stuff from an external storage device, meaning a nice increase in performance, and well, plenty of people still DO have a need for larger storage devices due to family photo albums, home movies, rental movies, music collections and so on and so forth. That IS a “commercial application of cheap hardware.”
Because it’s not really a real concern? Already you can fling files around on and off of these boards at speeds people already would find acceptable. I still meet people putting up with 10MB/sec transfer rates off of aniquated USB sticks.
In fact, as just pointed out: http://www.arndaleboard.org/wiki/index.php/Main_Page — that’s new as of this month and addresses every concern you have mentioned. There’s going to be a constant sea of these sorts of boards.
What I’m ‘defensive’ about is you honed in on the most trivial and inconsequential of details about a single board as if you were offering anything insightful about it, completely ignoring the actual conversation about closed drivers and the sort of opportunities they’re blocking. You made a non-comment and I called you on it. Congrats.
I already took a part on that conversation and I’ve done that multiple times before, including pointing out all the fascinating possibilities cheap, low-power ARM – boards offer.
No, you just seem offended about the fact that someone would dare to say anything even slightly negative about one of these boards.
I think you’re taking all of this way out of proportion. WereCatf and I have had multiple conversations in the past about alternative computing platforms and my understanding is that she is very much interested in the myriad technologies out there, especially ARM. However, anyone (you and I included) would be a fool to ignore the implications of closed hardware on that platform.
You might want to step back and try to get a little perspective before mouthing off and making a fool of yourself. Just a suggestion.
Though “terribly slow” probably goes too far? Maybe “slowish” …and OTOH quite enough for many other usages – reasonably comparable to 100 MBit Ethernet, when the little board is used to serving something over it.
PS. ( WRT http://www.osnews.com/permalink?543124 ) and with USB, CPU hardly idles when reading – USB is more CPU-intensive; maybe USB2.0-only is not such a bad thing on meagre CPUs.
Edited 2012-11-27 10:17 UTC
AFAIK, it is only USB 3 that introduced proper use of CPU interrupts for signalling, while previous releases required constant polling by the OS. Combining that with the larger packet size, a USB 3 stack might actually end up going easier those low-performance chips.
At least if OSs and devices supported it well, which, considering how well Linux 3.2 can react on dmesg and elsewhere when I plug something into my USB 3 ports, is far from a given.
Edited 2012-11-28 06:48 UTC
Number one the Mali 200 and 400 are quite primitive that Lima covers. Lima only covers Mali 200 and 400. Lima goes up to Opengl ES 2.0 max Mali 200 and 400 supports is Opengl ES 2.0 even with closed source drivers. So if Lima ever support Opengl ES 3.0 it would be ahead of all the closed source drivers in existence for that hardware.
Arm themselves directly publishes the source code for the Mali T6xx series on. This is was even strange to ARM when they did it. One of there competitor in enbeded video controllers texes instruments are the ones behind Lima. So it was kinda slap in face.
http://lists.x.org/archives/xorg-devel/2012-May/031250.html
Mali T6xx series does not have to be reversed. Lima mostly does not have to reverse 200 and 400 either. They can email ARM and get answers if something is not matching documentation.
Aristocracies something most people don’t know. It is important. 12 months ago it was officially announced that the Linux Frame-buffer driver system will die.
There are 3 video output stacks on Linux. DRI/DRM, V2L and Framebuffer. DRI/DRM and V2L are able to share buffers stable with each other. Framebuffer is Linux first video output system. Today at driver level is basically compatible with nothing else. DRI/DRM can provide Framebuffer syscall emulation.
So the location of Linux embedded graphical drivers and desktop graphical drivers being two different drivers is going to end.
Aristocracies the Mali 200 is able to handle a compositing desktop using wayland. So you don’t need super big video cards todo a compositing desktop.
There are other things that will upset people about Mali items like photoshop under windows off load to the gpu. Yes Mali 200 and 400 don’t have that power.
Really watch the Linux space carefully. The new driver system to replace framebuffer driver system and make it dri/drm compatible is highly interesting. There will be less drivers to support more hardware.
Embedded is highly evil gpu can be from one maker output chip can be from another and you have to make them talk with software.
Replying to myself just to throw in: http://www.element14.com/community/thread/20081?start=0&tstart=0 for fun.
An entire collection of people who have been banned from the RPi forums merely for discussing issues with the hardware and firmware. Liz Upton apparently has discovered the term ‘concern troll’ and has used it vigorously in maintaining their image on the forums.
The impression the RPi Foundation keeps giving me with their interactions with developers and users is that we’re not really the people they care about since we’re not going to go and give them a press release.
I’m all in favor of supporting other ARM SoC boards because while they may not be any more open than the VideoCore IV, at least you won’t be dealing with people more concerned with their image than with users having problems with their product. Probably also won’t see a self-congratulating press release lying about the nature of what it is you exactly open-sourced.
I’ve been enjoying working on the ODROID-X, admittedly $120 — but it crushes every other board I’ve played with in performance and still has GPIO pins and hobbyist support. I’ll be picking up an MK802III as well.
Really, these things are so cheap it’s fun to collect. I’m especially looking forward to getting a hobbyist version of: http://www.cnx-software.com/2012/03/07/samsung-exynos-5250-dual-cor…
Edited 2012-11-24 20:34 UTC
BTW, the hobbyist version of that is already out: http://www.arndaleboard.org/wiki/index.php/Main_Page
Wowzers that thing is a beast and it has SATA support. Colour me impressed!!!!!!!!!!!!!!!
It isn’t. Read the whole discussion (and several before) and you’ll see they just got tired of dealing with a troll.
Raspberry Pi is about as open source friendly as most off-the-shelf PCs. Yes, you need firmware (BIOS) to bootstrap your OS and, yes, you need a closed-source GPU driver if you need accelerated video. But other than that it is a fully open platform. There are plenty of things you can do with the device without touching GPU (frankly speaking, if you wanted to do something performance intensive you should avoid these devices anyway).
I also would like to have a fully-open ARM SoC but please be fair. RPi guys have already done more than their share of promoting open ARM systems. It is not their job to produce SoCs with open drivers.
To be fair, they basically just took the “nearest” SoC available – the Broadcom R&D bureau ( http://en.wikipedia.org/wiki/Alphamosaic ) responsible for RPi SoC is a stone’s throw away from RPi headquarters.
There are also some personal ties involved (one of RPi founders/directors working at Broadcom)
the other rapid thing is raspberry pi’s obsolescence
In the comments above, it looks like everybody is complaining about some defects of the rPi. It may not be perfect, but (1) nothing is perfect (2) it’s ^a‘not 32.88 VAT incl.
Cheap per unit, but consider how much money was spent in total; how many resources expanded… (wasted?)
Especially when it seems that there are comparably priced less limited options; just with worse PR.