Slashdot Mirror
Open Source Graphic Card Project Seeks Experts
An anonymous reader writes "Could this dream of many open source developers and users finally happen? A 100% open sourced graphic card with 3D support? Proper 3D card support for OpenBSD, NetBSD and other minority operating systems? A company named Tech Source will try to make it happen. You can download the preliminary specs for the card here (pdf). The project, though a commercial one, wants to become a true community project and encourages experts and everyone who have good ideas to add to the development process to join the mailing list. You can also sign a petition and tell how much you would be willing to pay for the final product."
Does it count if it's more than 2 weeks old?
If you'd read-up on this subject, you'd have seen that these folk *do* know their hardware.
They are also not being overly ambitious. While they expect to be able to develop a card which has 3D accelleration for desktop applications, they make no bold claims about gaming.
Indeed, this card is being designed as the ideal desktop-card for open-source systems with open-source drivers and firmware. Any gaming performance, while unlikely, should be treated as a bonus.
I have already pledged my intention to buy one of these cards just out of curiosity.
Ripping an new rectum in the fabric of spacetime.
I think the company would make a ton of money just making these as a reference platform and selling them to University students looking for a way to program their own GPU on the cheap for research purposes. Heck, Xilinx should do it themselves, and give all these students exposure to Xilinx parts (and their crappy design software) before they even find out who Altera is.
This project looks interesting. I'd sign on to help out, but this gets dangerously close to what my Day Job is, and I don't think my management would smile on my participation...
tech source makes graphics cards for sun microsystems computers, i've got a raptor in one of my ultrasparc10's. I'm sure they have some fabrication experience, just visit their website, they've got quite a few products.
If you read the mailing list archive, you'll see that what they are proposing is a card with simple, OpenGL compatible 3D. The interface will be PCI at first. My impression is that they have mini-ITX boards in mind. The last paragraph of your post is correct: they will probably target commodity Linux (and significantly, BSD) boxes.
I think that this is a great idea. Right now, if you want open source 3D, the only good hardware available is the Matrox G400/450/550 line, and that's over 5 years old. I bought my G450 in 1999 and am still using it quite happily, but I would certainly buy an open hardware card from Tech Source if this project comes to fruition.
As someone on OSNews posted, this project could be profitable for a small company even if it would be considered a flop by ATI or Nvidia.
StarOffice had thousands of man hours put into it before Sun bought it for millions of dollars and gave it away as Anti-Microsoft charity.
It would be impossible for the Open Source world by itself to build a replacement for MS Office in any reasonable timeframe.
Talk about "false logic." Open Office is doing pretty well because it has had a huge amount of time and money put into it over the years. By the way, it existed for many years as closed source before it became open source, even before Sun bought it.
http://en.wikipedia.org/wiki/StarOffice
And it's not anywhere near being ready to replace Microsoft Office, but I guess they've only had 10 years...
Now before I get modded down, I be to remind whoever might read this that what I am saying is FACT. - bogaboga
I am glancing at the specs and I have a couple thoughts.
The first is that these are respectable specs - providing you don't want to to any gaming.
I think that is a really important caveat. I know that every once in a while people get all excited because the usual suspects port there games to Linux - you know ID and Blizzard come to mind.
It is a good thing that these two companies do this, but it is a bad thing that there are really only two companies that do this with anything approaching reliability.
Thing is... a card with these specs, especially considering that it is a year if not more away from reality will never cut it for any sort of gaming. You are going to produce a card with 3D support that doesn't have the muscle to handle any 3d games that are produced.
If you are fine with that then there is nothing wrong with those specs. This card will be able to handle email, porn and movies as well as anything ATI produces.
My 2nd thought is a bit more practicle.
Actually there may not be anything practicle about it. Might just be wishful thinking really.
What about 3DFX? What about OPENGL?
Between the two things isn't half the work already done?
I know it might seem insane - nuts even, but back in the day 3dFX had some very respectible hardware. They didn't fail cause there stuff was poop, they failed cause they underestimated nVidia (which in turn underestimated ATI). The hardware is still out there, the code is still out there. It just isn't being utilized.
Would there be anything wrong with utilizing these old resources to achieve this goal?
The proposal itself does not look like it is going over well. Look at the comments on the Tech's proposal site and you will see flames about the card not holding up to the standards of gaming such that ATI and nVidia do. If they are able to find programmers and developers that are superb at their jobs, maybe even ex-ATI or ex-nVidians, the card has a great chance at surviving among its rivals. Furthermore, ATI never had good software to begin with [drivers, etc] compared to their hardware. Opensource has also always seemed to prevail in one aspect or another in comparison to closed source. the development is a great idea, and even if it does not take off right away, it is a step in the right direction...
_
Free 27" Sony WEGA TV
I suppose they'll have to either buy a lot of FPGA to get price reductions or wait until the price of those programmable chips come down. Add to this the price of DRAM, 250 MHz 3-channel DAC (which would not be on the FPGA), power converter, serial Flash for FPGA configuration and extra discretes. My guess on the final price... Easily more than 300$US. The enthousiast willing to contribute to the open-source project will also have to get some simulation and FPGA tools which Xilinx do not give away for free (it's like 1000$ or more for the FPGA Tools and even more for the simulator).
FPGA's are great for prototyping designs, but the expense associated with the EDA tools to develop a 1.5M gates FPGA will limit the contributors...
Nevertheless, its a cool project if it allows more people to learn how to code in VHDL/Verilog. It will find a niche in universities where students don't have to pay for EDA tools.
Cheers!
the price of the FPGA (xc3s1500) which is quoted at Avnet at 70$US for the 320 pins version and 115$US for the 676 pins version (in qty of 25-99)...
Well, first Avnet charges a markup to stay in business; and second, I presume Tech Source wants to sell more than 99 of these babies. Xilinx quotes the price of an xc3s1500 as "under $20" for quantities of 250 000. Pulling some numbers out of my ass, I presume that Tech Source is going to make something like 10 000 of these cards, so they would be getting a big volume discount - maybe not $20 each, but certainly way lower than $70.
No, OpenGL isn't exclusively 3D, but the parts of it being used by Apple are.
Also, Quartz Extreme is most definitely using 3D hardware acceleration. Regular "old" Quartz used before 10.2 was purely 2D based, but Quartz Extreme leverages your 3D accelerator to render the desktop on screen - acting like "Everything is a textured polygon."
Transparency and shadowing are not 3D features. In fact, there is little in OS X that requires 3D support, and OS X uses only a small part of OpenGL. So, a card with good 2D support, antialiasing, and transparency would be just great for desktop use. (Incidentally, while people keep talking about these features as OS X features, OS X wasn't the first to have them.)
A tiny software company who put it together "on their lap" literally, without any billion dollar investments a-la Microsoft. Never underestimate the power of a small group of highly educated, passionate individuals. I bet most of the work that goes into cutting-edge graphics chips is done by a team of 5-10 people. Can such a team be put together outside NVidia/ATI? You bet!
Much of what you're saying is true about reprogrammable (Xilinx, Altera) FPGAs, but this "open source video card" would be an excellent application for one-time-programmable (antifuse) FPGAs, like the chips made by Actel or QuickLogic. Once the FPGA code has been written, there shouldn't be any need for reprogrammability, and antifuse FPGAs are cheaper, faster, and have much better routing potential than reprogrammable FPGAs. In fact antifuse FPGAs can come very close to the speed of an ASIC.
I'm actually quite surprised that they're opting for a Xilinx FPGA here; I must be missing something. Is there any particular reason that reprogrammability is more important than cost and speed for this?
The card will be in constant production changes from the day it leaves the shop. But this doesn't mean just changes in drivers or even firmware. It will probably be desirable to completely change the structure of the chip itself between revisions. Afterall, this isn't a professional chip, its a project. Starting out with only a few (partly broken no doubt) features; building its way up to being a truely useful chip over several years. But if it wasn't re-programmable, you'd be stuck having to buy a new one every few monthes when they fix bugs in the hardware or add new functionality. FPGAs are no doubt the way to go to begin with. I'm sure once they get to a stage where they have a stable design with as much (working) functionality as they want, they'l make ASIC versions.
The problem that I see is that ATI Radeon = 9200 cards run Quake3 reasonably well and they already have open source drivers.
I already have one of them and probably wouldn't buy a worse card.
I hate to say it, but your logic is completely wrong. Although OpenGL is an open standard, current cards all have vendor specific extensions for handling such things as multiple texture blending and vertex/pixel shaders. Also, the effort required to write an OpenGL driver is significantly greater than writing a DirectX driver.
The Windows driver for a DirectX card is not that complex - and there are several available reference sources (3Dlabs, ATI). The highly complex drivers out there at the moment are very heavily optimized for a given card - speed sells. But the central core of the driver is simple, with almost all work handled by one entry point that takes command batches.
I ought to know - I'm the guy that designed the Windows kernel interface to the driver back in '97, and it's basically unchanged to this day.
What's your definition of Open Source?
r eadme.txt)
_ results.asp?strOSs=39&strTypes=DRV,ARC&ProductID=8 65&OSFullname=Linux*&submit=Go!)
From their license (ftp://aiedownload.intel.com/df-support/7485/ENG/
"LICENSE. You may copy the Software onto a single computer for
your personal, noncommercial use, and you may make one back-up
copy of the Software, subject to these conditions:
1. You may not copy, modify, rent, sell, distribute or transfer
any part of the Software, except as provided in this Agreement,
and you agree to prevent unauthorized copying of the Software.
2. You may not reverse engineer, decompile or disassemble the
Software.
3. You may not sublicense or permit simultaneous use of the
Software by more than one user.
4. The Software may contain the software or other property of
third-party suppliers, some of which may be identified in, and
licensed in accordance with, any enclosed "license.txt" file or
other text or file. "
(Link to License is here -- http://downloadfinder.intel.com/scripts-df/filter
Hmmm. Doesn't sound too Open Source to me.
ATI already supposed specs for their R2xx cards. So everything up to a ATI 9200 has accelerated 3d support under X.org using the standard radeon driver. You won't get speeds as fast as the ATI drivers and some things like texture compression aren't supported due to patents but it gives good performance for something like chromium b.s.u and tux racer.
Check out gatos.sourceforge.net for info on the open source video input/output support for ati cards.
"When you sit with a nice girl for two hours, it seems like two minutes. When you sit on a hot stove for two minutes, it
Tech Source is the company that supplies cards for our Solaris workstations. The driver quality is pretty decent, but we are only doing 2d. I would guess that they are fully capable of doing a good 3d card though. My only qustion would be price, as I think we paid $300 for our last Tech Source card, and it was a 64meg card.
"He's lost in a 'floyd hole"
The driver is also in the XFree/Xorg mainline under the usual OSS licence
Matrox used to offer full specs for their cards. After they released the G500 they had a hissy fit and took their toys home though, and now you can't download a register level spec from Matrox for love nor money.
Looking at the products of the company tech source I think that they are mostly looking into using the design byself. They are selling special high resolution displays and graphics cards for aviation and other critical applications. These markets have extremely stringent safety protocols, of which many can not be met by the highly complex GPUs of today. In addition the exotic resolutions may pose a problem.
Having a straight forward design suitable for an FPGA would enable them add additional fail safe mechanisms and to qualify more easily for these applications. Oh yes, and they get others to work on their products for free. They could use rad hardened FPGAs for the final implementation.
Spinning an ASIC on par with a comparable performance FPGA costs about $50,000 for one device. A 90nm FPGA can roughly compete with a .35 CMOS process but with the ASIC you have a MUCH longer design cycle an less features like on-chip SERDES and Delay-Locked Loops. Also, if you screw up the ASIC its another $50,000 gone. An FPGA is MUCH cheaper for low volume applications, like this card would be. Plus the design tools for a Spartan 3 are free.