After a failed Kickstarter back in 2013, the GPLGPU project has delivered a first version of its free (GPL v3) GPU.
As explicated by Hackaday, this FPGA based GPU does 3D over VGA and works with a standard PCI interface. This is not exactly a powerhouse, but it's a starting point for truly free GPU solutions, and an exciting news for everyone in the FLOSS community since truly free GPU are quite rare.
You can read the official announcement here.
This discussion has been archived.
No new comments can be posted.
First Release of the GPLGPU - A Truly Free GPU
|
Log In/Create an Account
| Top
| 35 comments
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(Score: 2, Funny) by Anonymous Coward on Wednesday August 20 2014, @12:59AM
Eat my pineapple!
(Score: 0) by Anonymous Coward on Wednesday August 20 2014, @01:47AM
Let's see...
Serious question: Does GPL GPU run Quake, a GPL game?
Actual quote from game: "Player1 eats Player2's pineapple"
Modded: (Score:-1, Troll)
Something is seriously wrong here.
(Score: 1, Insightful) by Anonymous Coward on Wednesday August 20 2014, @03:52AM
What do you expect from SoylentNews....SoylentNews is people and the people are ignorant as shit.
(Score: 0) by Anonymous Coward on Wednesday August 20 2014, @08:07AM
Eat my pineapple flavored frosty piss!
All your pineapples are belong to us!
(Score: 3, Funny) by epitaxial on Wednesday August 20 2014, @01:01AM
It has little to no documentation, costs $1200, runs slower than a 68040 CPU. How soon until GNU HURD is ported? Then things will really look up!
(Score: 3, Insightful) by Hairyfeet on Wednesday August 20 2014, @03:31AM
While you made a funny the reality is? Its not possible for this thing NOT to suck, because every form of compression/decompression is proprietary and patented up the wazoo. Why do you think "enemies" like Intel & AMD or AMD & Intel share patents? Because you couldn't make a functional CPU or GPU without 'em.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 0) by Anonymous Coward on Wednesday August 20 2014, @03:39AM
> Why do you think "enemies" like Intel & AMD or AMD & Intel share patents?
> Because you couldn't make a functional CPU or GPU without 'em.
Well that, and patent pools raise the barrier to entry. Even if your startup comes up with a brilliant new technique and gets a patent on it, they still can't compete because of the chance they will infringe on something else is so high.
Yet another example of how patents don't protect the little guy, they fuck em - and kill progress in the process.
(Score: 2, Interesting) by cesarb on Wednesday August 20 2014, @10:21AM
Even ETC2 [wikipedia.org]? As far as I know, it's free of patent mess, which is why it could be added as mandatory in OpenGL (unlike the known-problematic S3TC).
(Score: 1) by epitaxial on Wednesday August 20 2014, @04:19PM
So they made this slow overpriced GPU that can only play open source codecs? This is a solution in need of a problem.
(Score: 3, Interesting) by Hairyfeet on Wednesday August 20 2014, @06:54PM
Its actually even worse than that, its a slow overpriced GPU that can't even play Quake I which was quite playable on a Voodoo I card! So not only is it slow and overpriced it can't even compete with one of the very first separate GPUs ever released!
I've said it before and I'll say it again, if the only positive you have for your product is that it meets the very narrow GNU definition of "freedom"? YOUR PRODUCT WILL FAIL because there aren't enough users that will put politics above all and that is exactly what your product is, politics above all. For everybody else that $10 Geforce 8 or Radeon HD2400 will run rings around your product while being cheaper to boot.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 3, Interesting) by tibman on Wednesday August 20 2014, @07:01PM
I'll define the the problem : ) Every modern GPU is closed. Open source drivers exist but are typically reverse engineered. In no situation does the open source driver match the closed source version in features/speed/stability. If you want a completely open GPU then this is a solution.
This GPU is two decades behind in technology though. Not bad for a first release : )
SN won't survive on lurkers alone. Write comments.
(Score: 2) by Hairyfeet on Thursday August 21 2014, @05:44AM
So you just admitted the ONLY REASON for this thing to exist is a POLITICAL STANCE on drivers...yeah, remember openMoko? it WILL fail, it WILL be forgotten, because time after time we have seen if the only positive you have is a GNU defined version of "freedom"? You are dead meat, as there aren't enough hardcore GNU zealots to keep you afloat.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 2) by mtrycz on Thursday August 21 2014, @10:24AM
I really dislike how much people like you overstate their opinions.
Your own machine and the internet as you know it runs in a big part thanks to GNU utils, GNU compilers, GNU/Linux, and GNU/anything, except for the parts that the companies have been successful in locking down. How can you not realize that "their flavor of freedom" is what's making your own machine half-free.
This particular piece of hardware is certainly not usable for your enterprise yet, much like the lately discussed free BIOS, but these are the first steps, and give it time.
At the very least, leave it as is, instead of bashing it furiously. Nobody cares.
In capitalist America, ads view YOU!
(Score: 2) by Hairyfeet on Thursday August 21 2014, @01:29PM
Sigh, having trouble following a sentence? All those things you listed have other things going for them besides a GNU defined version of freedom...oh and they are all backed by corps, just FYI. And you can wave your flag all you want, I have history on my side because from OpenMoko to Pandora to the KDE tablet we have seen if the ONLY POSITIVE METRIC you have is a GNU version of freedom YOU WILL FAIL because there simply isn't enough GNU zealots to keep a project afloat....tell me Mr Righteous have YOU ordered one? Lets see the receipt...yeah thought so.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 2) by mtrycz on Thursday August 21 2014, @02:30PM
I have no interest in it currently, too busy with my own sht, but I welcome the effort of people working on it. Who knows, maybe one day we'll be blessed with a working piece of hardware that is also free.
If not, I'll still have welcomed the effort.
In capitalist America, ads view YOU!
(Score: 2) by Hairyfeet on Thursday August 21 2014, @08:39PM
Laugh out fucking loud,Mr Righteous ain't buying one...NO SHIT, that is because YOU DON'T NEED IT because obviously the shit you have now works or you wouldn't be able to respond to this post. I swear smug self righteousness among Linux zealots is more common than porn on the Internet, along with your hypocrisy as you trumpet how wonderful a thing is...that you will NEVER own.
Mark my words we will NEVER hear of this thing again except maybe an obituary, where a bunch of guys like you will talk about how wonderful an idea this thing you never would buy in a million years was...ROFL.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 2) by mtrycz on Friday August 22 2014, @08:29AM
Well, I STILL welcome the effort. Why are you being such an ass about it?
In capitalist America, ads view YOU!
(Score: 2) by tibman on Thursday August 21 2014, @01:13PM
If you say it is, okay. But for me it is the ability to have up-to-date drivers for new graphics hardware. Linux driver blobs are missing features that the windows version has and nobody can add them in. Maybe that is politics for you.
SN won't survive on lurkers alone. Write comments.
(Score: 5, Insightful) by cafebabe on Wednesday August 20 2014, @01:14AM
This is very impressive. Don't worry about the performance. I believe that FPGA competes at a 5x disadvantage to ASIC at the same scale. So, with minimal changes, if any, the current design should work at 1280*960 or more. That's more than enough for many practical applications, such as Point Of Sale terminals, Heads Up Displays, toys, embedded applications and budget tablets.
1702845791×2
(Score: 2, Insightful) by tftp on Wednesday August 20 2014, @03:23AM
Point of Sale terminals: they are not severely price-constrained; but still an ASIC GPU (embedded into the MCU) costs less than an FPGA and a Flash for it. In any case the MCU choice is <1% of the cost of the terminal. You'd probably also lose money on custom graphics libraries and drivers.
Heads-up displays are sensitive to power draw and size of the package. They will prefer ASIC over FPGA, especially because cost is not a concern. (In consumer products of high volume an ASIC would be cheaper too.)
Toys are extremely cost-sensitive. An FPGA is most likely not an option; an Artix (XC7A100T) costs well over $100, and it requires at least two or three power rails (1.0V, 2.5V for LVDS, and 3.3V.)
A budget tablet is built with COTS MCUs that have ARM for the CPU core and some simple GPU for the LVDS interface to the screen.
I am not sure what would be a good business case for using a GPU in an FPGA. I can understand that in principle this is a good thing to have, just as it is theoretically good to have a PDP-11 implementation in an FPGA. But in most practical applications FPGAs lose to custom ASICs, with exceptions in areas where frequent gateware updates are expected, or when the volume is too low to justify making an ASIC. There are many projects at OpenCores.org that are fun to work with, but probably not necessary for any practical implementation (like R-Pi, for example.)
(Score: 2) by maxwell demon on Wednesday August 20 2014, @05:24AM
I understood cafebabe as saying that the only thing you'd have to do to make this a viable GPU is to put the design with minimal changes on an ASIC. Which addresses exactly the point you make.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 1) by tftp on Wednesday August 20 2014, @06:02AM
I understood cafebabe as saying that the only thing you'd have to do to make this a viable GPU is to put the design with minimal changes on an ASIC.
I didn't get that impression. And perhaps it's not even possible/practical. Today's PCI interface allows you to make an ASIC, but it has to be on an add-on card for an old desktop computer, as parallel PCI busses are dead as a doornail.
The PCIe is more interesting, and it opens some ways to use this ASIC as a component in single-board computers... but the project does not have PCIe. Not that it's difficult to add, since PCI/32 and PCIe cores are (IIRC) compatible from the application end, but they require MGTs on the FPGA - and that costs dollars and watts. I don't know what PCIe cores are available for ASICs, as LogiCore netlists are built with Xilinx primitives.
Another bus that the developer is going to add is AXI [xilinx.com]. However this bus is specifically intended for use within an FPGA, in a soft SoC. I am not sure how would one use it as a standalone ASIC. Altera's Avalon is used with their soft CPUs (Nios.)
This explains why so many modern chipsets are integrating a reasonably good GPU. You do not want to mess with yet another bus in a low power design, as all this handholding and bit-shuffling costs you in size and in power. A GPU that is connected to the ARM via local AXI, and all that is implemented in one ASIC, along with GPIO and whatever else you need, is the ideal solution. Make it in millions, and it allows someone else to sell budget tablets for less than $100.
The GPL GPU as a standalone GPU chip is not very likely for these reasons. At very least it ought to have PCIe, otherwise you'd have to route full width busses, and that's a task that not even every PCB layout tool can do well. However that chip would have to compete with integrated solutions that are already working in every smartphone and every tablet on this planet.
As a chip it could be handy under control of Atmel, PIC and TI microcontrollers. Most of them do not have a GPU inside, and it is all but impossible to run a large LCD with bit-banging. However those MCUs do not have PCI to talk to this GPU design. One would need high speed SPI or something similar. Arduino runs the Ethernet interface over SPI just fine. Perhaps that's one of good niches for this GPU, as adding SPI and a local DRAM connection is trivial. An accelerated GPU does not require a huge bus bandwidth, and there are many industrial applications that need to draw colored boxes and lines.
An integrated CPU, GPU and GPIO, all under GPL, and manufactured as an ASIC, could be an interesting idea... but patents (?) on CPU command sets or some implementation methods may make that difficult. If successful, that would be an equivalent of the chipset (TI's OMAP, etc.) that drives a smartphone. It would be free, but it would require extreme investments ($10M) to implement.
(Score: 2) by Hairyfeet on Wednesday August 20 2014, @02:45PM
Sorry but it still wouldn't be viable as both AMD and PowerVR sell chips for industrial boards that cost less than $2 a chip that will run rings around this thing while using less power.
Lets face it the real problem with this "truly free" hardware idea is the ONLY selling point they have is GNU version of freedom...that's it. And time and time again we've seen if the ONLY selling point you have is GNU freedom you are dead meat, there just isn't enough folks who care THAT deeply about GNU freedom to make that a viable market.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 2) by Hairyfeet on Wednesday August 20 2014, @09:26AM
Uhhh X86 POS, HUDs and the like use ultra low cost chips like the PowerVR chips which cost a couple cents, no way you can compete there, and on the ARM front they likewise have GPUs that cost pennies a pop, no chance on that side of the fence.
Honestly if they sell 4 figures planetwide I'd be amazed as the ONLY ones that are gonna buy this are RMS levels of militant about the GPL, for everybody else there are MUCH cheaper solutions that run rings around this thing. I mean did you read TFA? It MIGHT run Quake I some day...maybe. That is a game that can run smooth on a $50 tablet with DOSBox and this chip doesn't even have the power to run it...yeah its DOA for all those that care about more than politics.
ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
(Score: 1, Interesting) by Anonymous Coward on Wednesday August 20 2014, @10:27AM
yeah its DOA for all those that care about more than politics.
Remember when they said this when rms made his initial announcement about a free operating system?
(Score: 1) by epitaxial on Wednesday August 20 2014, @04:21PM
Yeah its called HURD and nobody uses it. RMS does the userland tools which can run under any operating system.
(Score: 2) by meisterister on Wednesday August 20 2014, @01:58AM
If this GPU can be used for compute, then we're good to go. This looks like it would be the hardware hobbyist's dream: a free core that they could use not only to add 3d graphics to their creations, but they could also try their hand at implementing HSA.
(May or may not have been) Posted from my K6-2, Athlon XP, or Pentium I/II/III.
(Score: 1) by tonyPick on Wednesday August 20 2014, @07:14AM
but they could also try their hand at implementing HSA
Or they could just buy a Beagleboard-xM, which comes with an ARM-A8 plus a Ti c64x DSP core (plus a PowerVR GPU), for £150 or so (according to farnell) and much better documentation.
I mean, this is fun and all, and could lead to something awesome, but unless you really want to play with it specifically there are better alternatives for doing multi-core hacking.
(Score: 4, Insightful) by kaszz on Wednesday August 20 2014, @02:02AM
The general catch usually is that any sufficient large FPGA will require a synthesis tool that cost way more than your first borne. The second is that designing a multilayer circuit board at high frequencies ie 100s of MHz requires real hard attention to details and money (again).
And when it works, you will find something that doesn't but will require you to run this cycle again..
(Score: 1) by tftp on Wednesday August 20 2014, @04:06AM
You don't need Synplify Pro. Xilinx's Vivado WebPack [xilinx.com] is free:
This set of FPGAs should be sufficient for pretty much any need, from the low end (Artix) to the high end (Kintex.) And if you need one final synthesis done for production, find someone with a tool that you need and ask them to run a project for you. I'm sure you can find such people in many places. OpenCores would be, of course, the best hunting ground.
(Score: 2) by kaszz on Wednesday August 20 2014, @02:35PM
The problem is that the best chips that the free version allows you to use may not be sufficient enough. Which pushes people to use multiple chips which complicates design instead.
(Score: 3, Funny) by maxwell demon on Wednesday August 20 2014, @05:26AM
What does a first borne cost?
The Tao of math: The numbers you can count are not the real numbers.
(Score: 1) by pkrasimirov on Wednesday August 20 2014, @08:12AM
About 1000 sleepless nights, 2x superscalar CPUs (some can do with 1x but it's difficult), 2GB patience (8GB+ highly recommended for extended lifetime), and some amount of cash for bus (or other routing mechanism) etc. depending on where do you live. Good thing is once the project gets going it gets easier. However with maturity the maintenance needs rise overtime.
(Score: 1) by gardintrapp on Wednesday August 20 2014, @09:02AM
The problem is not so much the cost but the fact that you have to use a proprietary fitter (and most likely also the synthesis tool) to be able to use your free code. It is like having to use a proprietary compiler on free software.
(Score: 2) by kaszz on Wednesday August 20 2014, @02:13PM
Which points to the next problem. FPGAs are configured (programmed if you will) with very proprietary tools. And electronic circuit design tools with good integrated simulation is not free. Otoh, the original FPGA patents is from 1985..