Linked by Eugenia Loli on Sat 27th Nov 2004 21:01 UTC, submitted by Lars Roland
3D News, GL, DirectX A company named Tech Source has stepped up to create an open source 2D/3D graphics card. The project has an open mailing list and a spec proposal for the final card can be downloaded here. This project is a groundbreaking effort at creating truely open hardware, and is great chance for everyone who is interested in 2D/3D programming to see how a modern graphics card works.
Order by: Score:
v blurb
by Mark Brophy on Sat 27th Nov 2004 21:12 UTC
v RE: blurb
by Eugenia on Sat 27th Nov 2004 21:23 UTC
v Typo
by Anand on Sat 27th Nov 2004 21:26 UTC
v stop harrasing the editor
by bryce on Sat 27th Nov 2004 21:34 UTC
by Anonymous on Sat 27th Nov 2004 21:43 UTC

I'll buy it if I can get a hold of one when it's ready.

Awesome news!
by Damian on Sat 27th Nov 2004 21:45 UTC

This is great news for the open source world! Although it certainly might not compare with the latest and greatest of ATI/nVidia, given a reasonable price and support I'd definantly jump to purchase one. To me this is quite appealing since I am not a very big gamer but lack 3D driver support from ATI (freebsd). Maybe this will serve as a wakeup call to the big two into putting more efforts into open source.

Will be a commercial failure
by Anonymous on Sat 27th Nov 2004 21:56 UTC

Sorry, but this project will be a total flop with all but the ideological hard core of the free *nix community, and that hard core is a miniscule market.

When it comes to 2d, most people won't be willing to buy an open source graphics card for $100 when they could instead buy a used nVidia TNT2 (or similar) for $15. In 2d, the fact that full-featured drivers for this card can be included in free kernels will have little performance impact relative to the fully open but stripped-down 2d drivers available for (e.g.) nVidia cards.

In 3d, this card will be playing "nineteenth fiddle" to the market leaders--nineteenth, if it's lucky. Those few *nix users who *need* excellent 3d performance won't be able to use this open source card even if they are willing to pay a premium; it simply won't be fast enough.

by Anonymous on Sat 27th Nov 2004 22:10 UTC

RE: Awesome news!
Thats why I sold my Radeon 9800 Pro for an nVidia GeForce FX 5900.

RE: Will be a commercial failure
Yes I completely agree, it will have to be fast enough to play decent games on, and against companies such as nVidia and ATI they just won't have the funding to do what they want.

why i'll buy
by tech_user on Sat 27th Nov 2004 22:15 UTC

i'll buy because i need accelarated video decoding and scaling, and i need some opengl 3d handling. if this card does this - im happy.

i do not need fudged and compromised super-fast games-oriented graphics. (ps this is why 3d professionals don't use the high end consumer cards - because they cheat. they use professional lines whcih nvidia and ai will admit they produce).

by Anonymous on Sat 27th Nov 2004 22:19 UTC

wonderful news !!
after opensource , open hardware.

please hardware manufacturers/designers, consider that there is millions of potential buyers i the great blue planet.
why cant an manufacturer/designer create a tivo/pvr just like Dreambox, but for tv viewing/recording with harddrive,nic, dvd-recorder etc. it would sell thousens of boxes and let the opensource make the drivers/firmware/software

just like stated - itīs wonderful news

by Renaldo on Sat 27th Nov 2004 22:25 UTC

i do not need fudged and compromised super-fast games-oriented graphics. (ps this is why 3d professionals don't use the high end consumer cards - because they cheat. they use professional lines whcih nvidia and ai will admit they produce).

Yes, they use video cards that cost a lot more than the fastest consumer cards do. So what is your point? That this card will not only be inferior to the high-end gaming cards but also pale in comparison with the high-end workstation cards as well?

by Lars Roland on Sat 27th Nov 2004 22:36 UTC

This card is more in competition with Matrox and other 2D professionel cards, and the 3D performance can compete with those cards (and even some of the lov end 3D cards in production today). Also there is possibly coming dual head versions of the card (people tend to pay big money for dual dvi), and the card is produced on a small budget so Techsource does not have to sell millions of copies to get there investment home.

2D Perf
by Anonymous on Sat 27th Nov 2004 22:37 UTC

There's a market for a strong 2D card that would provide an accelerated desktop under *nix. nv drivers are awful as far as 2d perf as concerned (same for closed source nvidia). I would happily shell out 100 euros for a card that would run XRender and XComposite really fast. If this can be a startingpoint for a nice 3d card, then all the better, but a good 2d card alone would already be great.

What's a "flop"?
by Timothy Miller on Sat 27th Nov 2004 22:39 UTC

This card could be a flop by the standards of ATI and nVidia and still be wildly successful for the small company that is developing it. I'm just speculating at this point, but I doubt we'd need to sell more than 10,000 of the first run to make it viable.

by anon et al on Sat 27th Nov 2004 22:42 UTC

I've checked out the specs; except from 128 MB of memmory (which is great), I really do not see what makes this card any more attractive than the Matrox G400/G550 Series. These cards are already considered "open" and works on several different platforms without x86 legacy bios initialization. How can the TechSource card compete with Matrox? Matrox cards have some nice 2D hardware acceleration features like blitting (which is not a feature on this card?) The G400 also have a ramdac running at 360 MHz...

by Anonymous on Sat 27th Nov 2004 22:47 UTC

Just get the required feedback from the experts. possibly put this on the xorg mailing list and make sure you have a flexible base and room for wriggling stuff and you should be ok. if the world needs these kind of cards you should be able to sell enough to see profit.

I know quite a bunch of people have been waiting to see something like this. alteast it puts pressure on the competition

Effort worth trying
by Metic on Sat 27th Nov 2004 22:52 UTC

Even if they tried to produce closed source non-nVidia/Ati-based cards, could they compete with nVidia and Ati very successfully? Probably not, the competition is hard enough for current nVidia/Ati manufacturers not to mention independent players like Matrox. And normal MS Windows users would probably not be interested in the product anyway, whether closed soucre or open source.

Still I think that this project has some potential, just because they would be so much different from everyone else, and because there is a growing potential market for this kind of open source cards.

There are lots of people using systems that don't have any sort of (at least proper) 3D card support yet: NetBSD, OpenBSD, many smaller fast developing hobby operatings systems etc. There are thousands of such users that might be interested in this produt and lots of developers who have been eagerly waiting for something like this.

Besides, hardcore 3D gaming is not the only market for 3D. Many people would be perfectly happy just to have at least some very basic 3D support for their OS, maybe to use Blender, or to occasionally play Tuxracer etc.

If the large open source community could take part in the development of drivers from the start, maybe the price wouldn't become too high either?

by Lars Roland on Sat 27th Nov 2004 22:54 UTC

Please note that The spec for the card is not final, fell free to share your ideas on the maling list. There is also a petition online, to help Techsource size the market:

Not going to happen
by Dudley Do Wrong on Sat 27th Nov 2004 22:55 UTC

This kind of design spec would required ASICs and FPGAs wont cut it. The NRE costs to design and test will probably be too high for the number of customers willing to pay for the part.

Hmm, Neat idea.
by Mike on Sat 27th Nov 2004 22:58 UTC

I hope they do well. I like the idea of creating open hardware rather than cracking/reverse engineering the established cards.

However, as far as Linux stuff goes, Nvidia's drivers are already pretty good. It seems unlikely that someone who wants to play games in a Linux environment would choose Tech Source over Nvidia simply on the grounds of open source drivers.

I wonder if it wouldn't be more effective to just keep begging Nvidia to release the source code. This seems out of the question at the moment, but they have been more friendly than average and as more businesses take the plunge into OSS, Nvidia might be one of the companies more willing to make the jump.

At any rate, i wish this company well. If they can be competitive both performance-wise and cost-wise, i would buy a card in a second. But i'm not sure i would put my open-source ideology before my wallet in this matter. Maybe i'm part of the problem :-(.


Re: Matrox
by Lars Roland on Sat 27th Nov 2004 23:01 UTC

Matrox open ?, yes some of it is, but not anything near what this card will be and the Matrox Linux/*BSD support for there cards is non exsistend today. I used to be a die hard Matrox fan, but try running xcompmgr on a dual head Matrox and you will see what I mean. Also this card should (at least in a dual head version) be able to achive much higher resolutions than any of the well suported Matrox cards today (that is G400/G450/G550, the parhealia genereation has crapy Linux suport so even thoough it is a god card, I still consider it a no go).

Matrox is becoming less open
by Timothy Miller on Sat 27th Nov 2004 23:06 UTC

Matrox used to be open, but they've been backing off on that. I recall someone recently complaining that Matrox has taken their specs offline, so unless you have a friend who already downloaded them, you're out of luck. (I haven't bothered to verify this myself.)

Maybe they'll fail, but we will reap some rewards ...
by Anonymous on Sat 27th Nov 2004 23:10 UTC

If this card makes it to market and there is serious initial uptake, it may force ATI/nVidia to open up their cards. Think about it: they have life easy right now because Linux/*BSD users are willing to put up with crumby drivers. If there is serious competition on that front they may decide that it is best to open up a little now than a lot later. Because if somebody does get a toe hold in the Linux market (hardwarewise), they are going to want to go after the more lucrative Windows market.

by Anonymous on Sat 27th Nov 2004 23:22 UTC

"However, as far as Linux stuff goes, Nvidia's drivers are already pretty good. It seems unlikely that someone who wants to play games in a Linux environment would choose Tech Source over Nvidia simply on the grounds of open source drivers. "

they are big hassle because the drivers are not in tree. this is a huge problem we cannot ignore. look at any mainstream disto mailing list to see how much it is a issue for users

Lets hope this is the start of something great.
by NemesisBLK on Sat 27th Nov 2004 23:31 UTC

I wish this company much luck with their endeavor. At least there is a company out there willing to try a different route. Too bad they're not more of them out there. Yea they don't have the R&D and budgets of the Nvidias, ATIs, and Matroxs of the industry, but you gotta start from somewhere right? This card, if it makes it into production, will have a place in my box if it performs well.

RE: Re: Matrox
by Metic on Sat 27th Nov 2004 23:32 UTC

Yeah, Matrox drivers and support for older generation cards (up to G550) are still pretty good, even competitive with nVidia, but the support for newer and faster Parhelia-based cards is far from good: 2D works ok, but 3D support is rather awful, there's even no Linux kernel 2.6 support yet. And this is the situation with Linux (only Fedora/Redhat officially supported). There's no Matrox support nor drivers whatsoever for any BSD or other UNIX flavor.

Matrox has been promising to develop better Linux drivers for Parhelia cards, but those have been promises only for a long time (the last Linux driver for Parhelia cards is 1 year old already!). The sad truth is that the Matrox web forum for Linux users is full of unhappy customers nowadys: As a Matrox user I'm still hoping, but I would now certainly rather choose something like this new Tech Source card if it won't be all too expensive.

ATI and nVidia won't open....
by Timothy Miller on Sat 27th Nov 2004 23:33 UTC

There are two obstacles to nVidia and ATI opening:

(1) They have IP in their designs licensed from 3rd parties which they cannot legally publish.

(2) They want to hide implementation details from each other, because they compete with each other and do not want to give each other an advantage.

I have the advantage in that I have no encumbered IP. I designed my first GPU from scratch, and I'll do the same with this one. ATI and nVidia take shortcuts by buying IP, but then they get stuck with all sorts of legal restrictions.

Here's an example: At Tech Source, we declined to buy a PCI interface from another company because it would have legally prevented us from EVER designing our own (that's how the contract was worded, seriously!). We decided to tell them to take a flying leap, and then I designed my own from scratch. It wasn't THAT hard.

Oh, and then there's pride. nVidia swears up and down that they can't release specs because of 3rd party IP. If they change their minds, they'll prove themselves to be liars.

Nvidia design method
by John Blink on Sat 27th Nov 2004 23:40 UTC

I remember seeing a video at CNET showing how Nvidia designs video cards.

It was purely designed in software and then with a simple email that put forth an order to produce the chips.

If they are going for an open approach. They would be able to create open source tools for chip design, and design and simulate there chips purely in software.

I see this as a very long term project. It needs people who will committed to it, to make it less of a long term project.

by X on Sat 27th Nov 2004 23:43 UTC

This is a worthwhile project that I commend. Personally, I would buy any hardware component that will let my operating system (BSD/Linux) configure it and function properly with minimum fuss. I dream about the day when open source printers, modems and etc become the norm. There's a potential market for such hardware.

by Rayiner Hashem on Sat 27th Nov 2004 23:43 UTC

The card exits not so much for ideology, but because of practical issues. The X/DRI folks are severely hampered in what they can do by a lack of complete card specs. Not even ATI released all the specs for the R100 and R200 chips. Microsoft and Apple can do lots of experimentation with their GL drivers for stuff like Quartz Extreme and Avalon, because they've got the driver source, but the X and DRI folks are limited in that regard.

Thx to Tech Source
by Moshroum on Sat 27th Nov 2004 23:50 UTC

Yes, really big thx to tech source. I hope, that this project will have a happy end:
- create a stable card
- create a stable driver
- sell a lot of these cards (for a good price of course)
- get a big community
- show the 'big ones' what the open world can do

And dont forget to release date (should i mentione hurd at this place? ;) )

I would help if i could, but at the moment i began my first small microchip (so i am no really help), but maybe when the time of driver development comes......

Imagine if they can implement all requiered features in hardware for xcomposite, render and all stuff needed for eyecandy linux desktop.

Now do an alliance with a Linux Desktop Vendor (probably Novell, Lycoris, Walmart or Frys) and there you have it, a computer with as much eyecandy as an Apple ;)

BSD & other smaller operating systems
by Metic on Sun 28th Nov 2004 00:04 UTC

FreeBSD user might be happy with this card too. I don't know from my own experience, but I've read that the official nVidia FreeBSD drivers may often have quite much problems. Also other BSDs have practically no 3D card support/drivers, nVidia or any other, yet.

So unlike someone said above, you certainly wouldn't have to be some free software zealot to be interested in this card... Just consider how many thousands of BSD users there are. And, of course even more Linux and other OS (Solaris, Syllable, SkyOS, Haiku, Zeta, BeOS, QNX etc.) users might be interested in this product too. Also more people could then consider switching to BSD or to some other smaller OS if it had proper graphic card support.

by CaptainPinko on Sun 28th Nov 2004 00:10 UTC

this will lead to the birth of bsd/linux based consoles? hobbyist creating their own Nintendos? Combine this with GPL'd Quake 2/3 (eventually Doom 3) engines, VIA mini-ITX mother boards and this could become into a majour hobby!

Anyone submit this to /. ? That would definitely help get thiss project the publicity this project needs. Accept a few donations from those who support but don't need a card just yet and use it to offset prodcution costs (look at what PHK got) and I could see this really suceeding.

I've always wondered why nobody has done this yet. After all, there is an open source implementation of the SPARC around somewhere.

We need experts!
by Timothy Miller on Sun 28th Nov 2004 00:14 UTC

I would like to request that anyone reading here who have good ideas to add or can get in touch with experts on this topic, please join the mailing list and encourage others to do the same.

This product is truly a community project. The last thing I want to do is get my priorities out of order and put in something less useful in the place of something that people need more. Mind you, not everyone will agree on what is best, and I can only fit so much into the chip, but I think I can satisfy the broadest base of technical needs here.

by CaptainPinko on Sun 28th Nov 2004 00:17 UTC

read 2 of the links before posting and then went to the mailing list. Inore my last comment

by Timothy Miller on Sun 28th Nov 2004 00:20 UTC

There is no reason why we wouldn't sell chips to people who want to put them into embedded systems, such as game consoles.

Re: bah!
by Lars Roland on Sun 28th Nov 2004 00:26 UTC

Remember that the discussion on the list is just targeting the first card, if it is a succes, then the console idea might become a reality. So just throw in your 2 cents, on the list. The BSD stuff that is talked about in here, is a good idea, so if anyone in here knows any *BSD-centric forums/lists, then please spread the word about this project.

by poundsmack on Sun 28th Nov 2004 00:41 UTC

with some sort of corperate backing this might actualy work. i wish the best.

RE: perhaps...
by NemesisBLK on Sun 28th Nov 2004 00:49 UTC

"Anyone submit this to /. ?"

It has...about a month ago.

Re: RE: perhaps...
by Lars Roland on Sun 28th Nov 2004 00:59 UTC

Yep we have been on slasdot once, but getting it up there once more, would be nice. It will put some new light on the development of this project. We are moving fast and a much has happend since the first turn on slashdot.

by The Lone OSer on Sun 28th Nov 2004 01:57 UTC

I don't see HOW this can work?, pretty much every aspect of
video card production is bound to be tangled up in patents which requires licenses which requires vast sums of money.
I know Matrox have EMBM, thats why nVidia and ATI pay for that license, and I know since MS brought a whole heap of patents off SGI, MS pretty much have the nuts and bolts of 3D rendering in their IP books.
Then there are bound to be a gazillion patents involving 2D as well; palette managment, windowing, bank switching, etc. etc. etc.
Also, if this DOES work, maybe an opensource PPC Mobo should come next ;)

I love this idea
by Chris Dunphy on Sun 28th Nov 2004 01:57 UTC

I'd ditch my nVidia card for a card like this, as long as my dual head worked, and I could do basic OpenGL for eye candy and my pretty OpenGL screensavers, and use programs like Celestia without too much hassle. I am not a big gamer, but I do need some 3D goodness...

I hope this project is a success, it is a cool idea.

I'll buy
by i3x171um on Sun 28th Nov 2004 01:59 UTC

Show me some fair results and I'll definately buy--seriously. I need a new video card anyway.

Re: I love this idea
by Lars Roland on Sun 28th Nov 2004 02:16 UTC

3D eye candy will work (as will quake3, and other games from that time). We have spcific keept track at all the spiffy 3D stuff that the xorg/enlightenment people could use now and in the future and made sure that the 3D engine will be able to do them.

RE:The Lone OSer (IP: 203.96.140.---)
by BR on Sun 28th Nov 2004 02:22 UTC

Sadly you bring up a good point. They could purchase the PowerVR Kyro, and Tseng patents, and build from there. A good 2D with an underappreciated 3D (IMHO)

by Alien on Sun 28th Nov 2004 02:49 UTC

I like the idea, and would support the company. It may not be the fastest out of the box, but maybe be more stable, and who knows, its new, and a new idea. Let time tell if its a good idea. I hate being skepitkal, it stifles my creativity.

Don't, if you know what's good for you
by mario on Sun 28th Nov 2004 03:00 UTC

I predict this project (provided you manage to pull this off to a stage where you sell stuff) to be a massive failure.

Very few people are going to pay US $100 for a card performing like an S3 64 (with more memory), but I doubt that you will be able to sell it at that price even. Expecially a year from now. I think you could sell maybe 500 of these cards, if you're lucky.

The ironic thing is, I actually like this idea a lot: being able to work with a PC graphics card (did once with an S-100 card) in assembler, making my own drivers etc., that woulbe be great fun.

OK, now feel free to be mad at me, but don't do that publicly if you plan on making a series of 500 cards, partially based on this post (and then end up selling even less).

by Anonymous on Sun 28th Nov 2004 03:10 UTC

I don't see why it'd cost so much to fab one of these cards. Video cards are only as expensive as you make them - modern Radeons and Gforces are pricy because they use very high-end RAM plus GPUs on the same level of complexity as modern CPUs. Something like a modem is dirt cheap because it is so simple.

A GPU as complicated as the one on the G400 would be relatively easier to fab thesedays, and still provide good performance for most uses. A reasonably priced, stable card that is assured to work with any OS would sell.

This company also has the ability to be more experimental with their cards than market conditions dictate to the big makers.

by mario on Sun 28th Nov 2004 03:13 UTC

Are you out of your mind? Do you have any idea how much does that COST?

Besides, they already said they'd be using an FPGA, which is not too cheap, either - but certainly a few orders of magnitude cheaper than have 500-1000 ICs fabbed.

RE: Don't, if you know what's
by X on Sun 28th Nov 2004 03:17 UTC

Very few people are going to pay US $100 for a card performing like an S3 64 (with more memory), but I doubt that you will be able to sell it at that price even. Expecially a year from now. I think you could sell maybe 500 of these cards, if you're lucky.

You might be right. But, that price applies to the initial stages. When a new hardware component debuts, it usually costs more, but gets cheap in the long run. In the long run, this open source card could costs $20 or less. For instance, a DVD writer cost $150 or more 1-1.5 year ago, but now costs as little as $30 (8x at Fry's).

Re: FAB???
by Anonymous on Sun 28th Nov 2004 03:30 UTC

Mario, this company *already sells video cards*.

you have to start somewhere.
by Anonymous on Sun 28th Nov 2004 03:39 UTC

There is a hole in the market, which none of the 'big players' are filling.

There simply isn't a graphics card available that supports all the features needed for a modern X-Windows desktop that has open source drivers, and/or a future development path.

There is something deeply unsatisfyign about this situation to many of us who are interested in knowing about how our computers work, as opposed to merely 'using' them - or being 'software consumers'.

So please shut up about how it 'won't sell', and 'won't perform' and 'can't be profitable' and leave those who are actually going to do the work to get it done.

All I know is that it is a damn sight better to try to do this, than it is to sit on your ass and resign yourself to a future where Linux's future on the desktop rests with whoever eventually gains monopoly control (and it's a race between NVidia and ATI at the moment) over the graphics accelerator market.

Gee, i dunno, maybe the project might even be educational, insprirational and fun?

But whatever - close your mind, accept the conventional wisdom thats been shoved down your throat, and continue to believe that the only important aspect of computing is the framerate which can be acheived in Doom3, nobody working on this project is going to stop you.

this company *already sells video cards*.
by mario on Sun 28th Nov 2004 03:57 UTC

They are still going to use an FPGA, not a fabbed IC. I wonder why....

Re: @ Anonymous (IP:
by Giovani on Sun 28th Nov 2004 04:17 UTC

so was Linux 10 years ago. think a bit. Where will we be in 5 years if we encourage such happenings! Don't be a moron. If you need the power go get an nVidia or ATI today. The rest of us who see the light we'll buy into free spec and open spec hardware. This is WAY better for everybody because it encourages COMPETITON (to spell it out for you competition is good for you the CONSUMER).

by punkass on Sun 28th Nov 2004 04:43 UTC

Sounds like a great idea. I'm definitely going to keep my eye on this.

A few questions
by Shurik on Sun 28th Nov 2004 04:50 UTC

First, I wish you the best of luck - a card like yours (PCI, lots of memory, good modern X driver, high resolution) would be wonderful for those of us trying to run a 2-3 monitor setup. However, I have several questions.

1. What OpenGL version are you targeting? 1.0 ? 1.4? 1.5 ? 2.0;) ?
2. And on that subject: OpenGL extensions. Could you, for instance, use the NV_* extensions or are those under a proprietary license?
3. XvMC ?
4. I suspect most people that would buy your card are looking mostly for good 2D performance, but your design is for a mostly 3D card. Am I misunderstanding?

Other applications
by justhe on Sun 28th Nov 2004 05:04 UTC

This project is pretty cool on many levels. I think there might be quite a few people interested in buying a relatively cheap pci card with 128MB mem and an FPGA on it, think about it. Also, as others have said, this is perfect to accelerate all the new XORG eye candy, it leaves the door wide open for all types of optimizations. I hope this succeeds.

HW specs
by xlnx on Sun 28th Nov 2004 05:14 UTC

Browsed through hw spec available on their site. Brave ones, I would say. Trying to push poor 3S to do PCI and DDR plus a bunch of block manipulation would just eat out everything that little baby has, not even talking about speed. This smells like 2V at least, which is otherwise fine, but bites on price. No free lunch, folks.

a very big mistake
by Osho on Sun 28th Nov 2004 05:35 UTC

Whoever is doing this isn't in their right state of mind. Open source concepts and advantages simply do not apply to hardware. The reason is very simple: To make a copy of software so that one more user can use it - you just have to download it. To make a copy of hardware so that one more user can use it - you have to actually manufacture a piece of hardware. This fundamental difference just makes it impossible to realistically have a really open source/open specifications hardware. There is a reason why none of the open source hardwares at have never been as successful as open-source software.


re: a very big mistake
by Shurik on Sun 28th Nov 2004 06:03 UTC

Well, there are a few advantages to open-source hardware. First, you can tap the skills of the community (they might not help you implement the thing, but at least they will give lots of advice and feedback). Second, you don't need to worry about ever writing any drivers, so you don't need to hire programmers.
I suspect that the reason most of the projects are unsucessfull is simply because they suck. There aren't many skilled hardware folk out there who have time to contribute. This is why an opencores project put together by a couple of students won't be a success (at until a few years' worth of improvements), but something led by an *actual hardware company* (e.g. Tech Source) just might make it.

Don't be so pessimistic!
by Dave on Sun 28th Nov 2004 06:53 UTC

To all of the naysayers... a lot of people were saying the same thing about Linux/open-source. "They'll never be able to compete with the big guys." It's a start.

Many Eyes
by jacks4u on Sun 28th Nov 2004 06:53 UTC

I like the open source idea, mostly because Many Eyes looking at an issue will find the best way to solve it.

I would buy this card, and applaud your efforts. I also feel you should consult the with the X developers, and perhaps enlist their help in the early stages.

NTSC output
by Matt on Sun 28th Nov 2004 07:33 UTC

Good, real NTSC output. The ability to output a true NTSC 480i along with HDTV 720p/1080i with out having to go through any scaling. This could be done with a DVI-I adapter similar to what ATI sells, but should also include S-video / composite output. The ability to program in accelerated video decoding would be great, similar to the MPEG-2 acceleration that exists in most other cards, but the ability to extend it to Xvid / H.264 / etc. Finally, the ability to input video instead of output. Combined with the video decoding (and possibly encoding) features from above and the ability to do NTSC / HDTV / DVI / VGA input could make this card a dream card for the HTPC market. On this front having a simple two line RCA input on the card for audio input would be extremely helpful.

by bob on Sun 28th Nov 2004 07:59 UTC

This card has to have the best drivers out there. Lots of other small companies have made kick ass cards but the drivers they made held them back.

If this card can sell for around $40 US you may be able to make enough money for a second generation of cards.

i would change the tech specs to these:

64mb memory (faster) + 4 pixel pipelines running at 185 mhz. (= ATI 9600 running at a lower speed - some 3d gaming features)

Brilliant, and about time
by Bryan Kagnime on Sun 28th Nov 2004 08:23 UTC

I don't really care so much for the 3d gaming aspect, distribute with the card an opensource operating system like Slackware with some 2d desktop eyecandy (translucency/transparency/openGL) and I'll buy a card for everyone I know with a comp. This'll show users *what* linux is all about, distrobuting a superior product and opening the market share for innovators.

ok, this is really bloody bugging me
by Vulpes Velox on Sun 28th Nov 2004 08:54 UTC

Why must something like this specifically be for linux like many of the posters are acting like?

Something like this is totally awesome for more than just linux. Nearly any OS with out good commercial driver support would benefit from something like this.

by Pier Luigi Fiorini on Sun 28th Nov 2004 09:22 UTC

I would see good DirectFB and Xorg drivers for this graphics card, I own a Radeon 9000 but I feel I cannot use all its power on Linux.

Re: Vulpes Velox
by Anonymous on Sun 28th Nov 2004 09:22 UTC

Actually the developers are supporting the sun framebuffer interface as well, and the standard VGA interface for DOS etc.

Given that the hardware, firmware and reference drivers will be completely open, this makes a pretty good starting point for alternate OS support to be provided.

There are simply more people who develop linux hardware, linux drivers and are interested in products for linux than there are for other platforms in the 'alternative OS' communities, and as such, Linux tends to be the development platform of choice.

However, if you would like to write a driver for your personal favourite OS to support this card, i'm sure the project would welcome your contribution.

helsinky university im developing...
by Anonymous on Sun 28th Nov 2004 09:42 UTC

Go for it, no one can say it will be a failure, even a company like microsoft almost intouchable is fearing free software, this can be the core to other proyects. The open source programming model is succesful millions of enthusiastic hardware developers are willing to compete even with nvidia, I will sure give it a try. Open Hardware, the final frontier...

by Roger Wolff on Sun 28th Nov 2004 10:28 UTC

This is exactly what I'm looking for....

As long as it can be configured as a Xilinx development board with PCI, onboard RAM and a video out.....

lets see wht happens
by Imran J on Sun 28th Nov 2004 10:55 UTC

THis is a interesting development, a company actually willing to develop something open, who knows that this product and firm may get corporate backing and things fly from their.

It a wait and watch approach, dont stop your life for now.

Go for 2D and good feature support
by U. S. on Sun 28th Nov 2004 11:01 UTC

Having rudimentary OpenGL support sure would be nice, but you will never be able to compete with nvidia and ati. Therefore please don't overdo it! There's no reason targeting the gamer audience. First of all, it's a minority which always needs more power. And if you are a serious gamer you need Windows anyway.

IMHO the prioroties should be:
1. Stable Hardware
2. Stable Drivers
3. Moderate Price
4. Passive Cooling (please! no need to produce a 500MHz GPU!)
5. Kick Ass 2D support (having all the X-Stuff properly accelerated)
6. Finally a card, that has all the features you would like in one card -- and they actuallu work under X! I'm talking about Dual-Head, S-Video out, perhaps Composite-In?

Once these features have setteled in the first or second batch of cards, one can think about accelerating the 3D engine to be able to run more than just gl-screensavers ;)

However, if the 2D signal quality sucks, this card will never make it. Realize that you are targeting not the average Joe who has learned to look for MHz/GHz anywhere, but computer savvy folks, looking for quality, efficency and feature support.

by Sebastian on Sun 28th Nov 2004 11:28 UTC

Sorry, I haven't read all the comments, but to all the "would pay 100$"-sayers on the first pages: I would be very surprised if a custom build FPGA based card with that specs produced in very low quantity will cost much less then 500+ $. And if everything works out well it might even have 50-60% of the perfomrnace of my good old Voodoo3.

Nonetheless it's a very cool project ;)

Regarding Earlier Comments...
by digitaleon on Sun 28th Nov 2004 12:48 UTC

From: U. S.
Subject: Go for 2D and good feature support
Having rudimentary OpenGL support sure would be nice, but you will never be able to compete with nvidia and ati. Therefore please don't overdo it!

I agree with the spirit of what you're saying here, but if you look at the kinds of desktop features -- and desktop eye-candy -- that are coming along over the next few years, decent OpenGL support is rapidly becoming a necessity. Even if you never play a 3D-accelerated game at all.

From: Vulpes Velox
Subject: ok, this is really bloody bugging me
Why must something like this specifically be for linux like many of the posters are acting like? Something like this is totally awesome for more than just linux. Nearly any OS with out good commercial driver support would benefit from something like this.

...which, realistically, is every OS that isn't a recent Windows or Mac OS X version. It's a disappointing situation, but not surprising given both nVIDIA and ATi are dedicating their resources to being nimble with acceleration performance and rendering features, lest they lose out to the other and end up the next Creative.

You have a good point. As I understand it, WalMart is marketing 'cheap' hardware 'appliances' running Linspire (or is my news out of date?). I'm sure they, and whoever helps them out with manufacturing/support, would be ecstatic about a video card with fully supportable graphics acceleration and rendering features. Such a deal is not beyond the realm of possibility (note: that's speculation only).

From: Osho
Subject: a very big mistake
Whoever is doing this isn't in their right state of mind. Open source concepts and advantages simply do not apply to hardware.

No-one's suggesting there won't be snags. Video cards are one of the few "PC Hardware" areas left for which there is really no 'industry standard' for interfacing with the cards. That's surely going to take time and effort to change, and more than 'sweat equity'. But this project, beyond being an exercise in open hardware, should help remove one more obstacle to genuine OS competition across microcomputers if it makes it to fruition. Granted, there are many more to go, but as was written in the Open-BeOS (now Haiku) FAQ previously, How do you eat an elephant? One spoonful at a time.

by Eddie303 on Sun 28th Nov 2004 13:19 UTC

The guy, who started the idea is an engineer, and as you know, the company makes professional graphics cards for Sun workstations. nVidia will not release the source, because it has patents, and it cannot afford to open.

Great Idea!!!!
by Anon on Sun 28th Nov 2004 13:19 UTC

This is actually a really good idea, but not really just for having a fully accessable 3d graphics system for *nix systems....i saw an article on slashdot about 6/7 months ago about shifting certain processing tasks to the videocard and how a large performace increase could be gained, one of the probblems cited in the article was that the cards had not been designed with this in mind and how wonderfull it would be if Nvidia etc would take notice and change a few things in the operations the boards could do.

After all a 3d graphics card is justa very good card at proccessing floating point streams isn't it????

by kaiwai on Sun 28th Nov 2004 14:09 UTC

IMHO they should have gone for the MAJC chip for the GPU; it is currently used in the SUN Visualization servers, and the GPU is licensed under a pretty liberal and open license; maybe the developers should look there before re-inventing the wheel for the umpteenth time.

Ram + Cost
by Alastair Boyanich on Sun 28th Nov 2004 15:16 UTC

I'd love to see the production cost driven lower by providing a card with zero MB (no memory) on it, merely a DDR or DDR2 slot. Should keep the cost down for producer, and owners can add 128..1024mb dimms. Part of my doctoral thesis's project's output is opengl 3d imagery that really benefits from 512mb+ ram.. cost an arm and a left nut for a card.

Also.. Alpha + Mac people, possible audience, also it'd be king of someone wrote 2d (at least initially) win32 drivers. I'd love to buy a couple and know a fair few people in the uni computer club I'll be telling to watch this project. I really really hope that it comes to market.

by Morin on Sun 28th Nov 2004 15:20 UTC

This is definitely a great step forward. Maybe it opens some minds towards what cooperation is about. Sadly I'd buy that card for fun and educational purposes only, and $100 isn't quite cheap for that. Anyways I'll keep it in mind, maybe the second generation is more for me.

> Open Hardware, the final frontier...

Huh? Complete sharing of knowledge would be the final frontier, if it is to come some day. You seem to be too focused on computers...

Hello Dev's.

I'm an electrical engineer with 7 years design experience (yes I'm young,) with a few questions for you.

1) who is your fab for these chips?
2) where will your end product card be assembled and by whom?
3) Who will be you ram supplier(s)?
4) what type of ram will you be using?
5) Are VGA or DVI or both, or multihead connectors planned?
6) Who is your chosen target market?
7) What problem are you solving for your market that is currently unmet?
8) Will you spin CLPD's, ASICs or FPGAs (obviously FPGAs are MUCH more money but are truely open and reconfigurable?)
9) Will you be developing in Verilog, VHDL or some other language?
10) Will you submit the completed design to FreeIPcores or a similar site?
11) What sort of license will be applied to the end product?

One last HOPE from me: enabling multi-monitor gaming for modern games in a much better way than Parhelia does.


by Mike on Sun 28th Nov 2004 15:40 UTC

The Eyecandy some of you are referring to simply doesn't exist yet. It's not a hardware problem, just no one has written the code. Yeah, if you really want it you can get drop shadows and some translucency (Xfce is especially nice). But at the moment, OSX style window animation isn't in any CVS' i'm aware of.

And when that code is written, it will probably work best on an Nvidia card with RenderAccel enabled.

I would love to see this company do well. Even if they aren't "competitive", they could still be successful. But I fail to see how they will provide any "eye candy" we don't already have access to. Nvidia's drivers have a perfectly acceptable openGL implementation after all.


Memory slots, questions
by Timothy Miller on Sun 28th Nov 2004 16:31 UTC

Lots of people ask lots of questions, many of which are answered in the spec, and many of which would be best asked on the mailing list. Perhaps we can reference them as part of our FAQ development.

Also, just as a note, it would be more expensive and potentially less reliable to put EMPTY RAM slots on the board than to solder 128MB of chips directly onto the board.

faillure? noway!
by webmind on Sun 28th Nov 2004 16:54 UTC

I'd buy one if it's not -too- expensive..
and making hardware for small userbases is possible and doable.. as can been seen with individual computers

Put 4 Ram slots on this card! Make it speed/size selecable.

That will save a huge amount of money on the production side, by not having to pick a chip and solder it on. Use a standard kingston module as a benchmark, and allow open source testers to post feedback on other brands/speeds.

A 512mg video card would sell! Upgradable to 2048mg, would sell the gamers and 3d modelers. That would be awesome for photoshop and video editing! With 7megapixel cameras coming out, this would be a major selling point.

RE: Memory Sell
by Mike on Sun 28th Nov 2004 19:04 UTC

yeah.... i'm not so sure you know what you're talking about. At all.

by AdamW on Sun 28th Nov 2004 19:40 UTC

nope, that's not going to happen - PowerVR are still in business and using the Kyro technology, actually. They aim it at cellphones these days.

by AdamW on Sun 28th Nov 2004 19:51 UTC

yup. I don't quite get the point here - the point of the memory on a graphics card is that it's very very fast. I don't see any benefit to putting a standard RAM interface on the board, because then you may as well just offload the flipping gfx processing to system RAM (as many integrated graphics solutions already do). If you're talking about designing some kind of custom memory interface for the card which will allow it to run as fast as soldered-on memory...heck, this thing is looking expensive enough to build already.

Re: BSD & other smaller operating systems@Metic
by Hagge on Sun 28th Nov 2004 20:36 UTC

"So unlike someone said above, you certainly wouldn't have to be some free software zealot to be interested in this card... Just consider how many thousands of BSD users there are."

What? ;)
BSD is _MORE_ free than even Linux (however any work made upon it doesn't have to be free, that's why i consider it even more free, do whatever you like.)

Anyway, I've had no problems at all with nvidia drivers for freebsd, however they don't work in netbsd or openbsd, which is sad. And yes, this would be great for those, for morphos, for aros, for skyos, for syllable, for any yet-to-come-os, for zeta, for ... If only it's affordable/fast enough it would be kick ass.

by Mike on Sun 28th Nov 2004 21:46 UTC

The "openness" of the card would benefit smaller projects than Linux. Linux already has openGL drivers for just about every Nvidia card and soon ATI will have drivers for the entire Radeon series. This card will be much more beneficial to the projects not yet big enough for the bigboys to help out.

RE: Go for 2D and good feature support
by Metic on Sun 28th Nov 2004 23:09 UTC

A good list. I mostly agree with the preferences you had there, especially with the good 2D quality and the fanless, quiet design.

Good 2D quality is most essential. Only long after it comes 3D performance (what ever some 3D fans might say about the importance of 3D even in normal desktop usage in the future). Also even only very basic OpenGL support could be enough at least for me but I am not really willing to use any graphics card with low quality 2D.

Later Tech Source might perhaps target the gaming market too but only much later, if they first succeed providing a good card for typical office type of work. It is dangeruous to be too ambitious in the beginning in this kind of a project. Start small, and then grow, that is a wise strategy (remember the beginnings of Linux, or GNU/Hurd on the other hand?)

Also, if you're a harcore 3D gamer, you might be wise to use a game console instead of a PC anyway, a good game console is much better and cheaper tool for that hobby.

RE:Metic (IP:
by BR on Mon 29th Nov 2004 01:11 UTC

"Good 2D quality is most essential."

I've noted that Tech Source already has a good 2D card with Linux support. I assume the chip used is under a NDA, and that's why it can't be used.

"Also, if you're a harcore 3D gamer, you might be wise to use a game console instead of a PC anyway, a good game console is much better and cheaper tool for that hobby."

Let's not open that can of worms.

Apple hardware.
by John Blink on Mon 29th Nov 2004 03:33 UTC

Will this work on Apple hardware. Or will it have to be modified too much for it to work.

Anything you want to make it work on.
by Timothy Miller on Mon 29th Nov 2004 03:46 UTC

Someone will have to develop drivers and firmware for Apple to make it work in a Mac, but the hardware won't have to be changed in any way. I am very interested in providing Mac support, although I don't yet have the means to for that, so I'm hoping some Mac developers will come along and help out.

by John Blink on Mon 29th Nov 2004 04:02 UTC


find a motherboard manufacture and embed it in
by bbbush on Mon 29th Nov 2004 04:53 UTC

since most people buy integrated motherboard with every function module on it! If only Intel would take it....

surely it will be a challenge to both design and marketing

good idea
by Robocoastie on Mon 29th Nov 2004 05:13 UTC

I've followed this news for a while now and think it's a great idea. The market could use another graphics card vendor. Does it need to be GeForce 6800 qualilty? I doubt it, that might not be attainable as it would depend on how much capital they have to engineer the chips.

by Anonymous on Mon 29th Nov 2004 06:03 UTC

If it can play doom3/et I'll by it in a hart beat. I'll probably buy it any ways since it must certainly will out perform my Radeon 7000

I don't get it....
by Jayclark on Mon 29th Nov 2004 06:42 UTC

For the people who say there going fail, thats just not very helpful. But I don't see why everyone screams for open soruce. An os yes. But Nvidia drivers always worked just as good on Linux as they have on Windows for me. Maybe I'm consider the "new users of Linux" that don't really care to have everything opens sourced. Will they make it? I guess they will by a small margin. But the majority of the market can care less about being open sourced. I mean If I see open sourced next to a close sourced hardware/software I'm not going to pick on that alone. But which gives better performace etc. And I doubt this will give better than Nvidia or ATI once they get their drivers up to par. Even if it does. They'll have a lot of ground to try to cover. I say go for it. You got to take chances every now and again. I'll buy it anyway and test it out.

RE: Why this can be great product
by Anonymous on Mon 29th Nov 2004 06:55 UTC

"Now do an alliance with a Linux Desktop Vendor (probably Novell, Lycoris, Walmart or Frys) and there you have it, a computer with as much eyecandy as an Apple ;) "

We already have that. Its called Windows, and already has a huge selection of good software and hardware, something that Linux doesnt have.

RE: I don't get it....
by Anonymous on Mon 29th Nov 2004 07:41 UTC

The reason why this is good, is because it's a step forward desktop integration for FOSS OSes. If this kind of project can pass the first round (and i'll surelly put off my wallet to help that), this could become something very valuable to the community.

Despite the quality of nvidia's drivers (which are really good), new users often have problems configuring 3D acceleration in X and this is slowing adoption of free solutions.

IMO, the point is to go for 2D big time, at least for the first generation (because there of course will be a second one ;) . Things like composite (which i believe only nvidia's aenderAccel givers honorable perfs) and damage should definitily get attention, and most importantly, everything related to Xorg. There of course should be 3D capabilities, and playing games is also important for adoption i think but don't even try to mimic the giant because you feel stupid about 3D and gaming capabilities, this is for the second round. ;)

They also souldn't try to compete w/ other GPU manufacturers for the first cards on windows, only focus on FOSS OSes, but i believe this is what's gonna happen (focus on foss).

I wish the very best to this project and the future of linux graphics.

Dual boot, dual cards
by John Blink on Mon 29th Nov 2004 08:08 UTC

Isn't it possible for us dual booters who already have an ATI or NVIDIA to choose which card will be default in the OS?

If true then, in Linux or other non-MS OS will be get adequate 2D/3D support whereas in windows we can still use our NVIDIA/ATI cards for gaming.

After a number of generations this OSS card could be really good for 3D workstations.

BTW is there a time frame for a first release? Or are we waiting for a development community to come around to this.

As an engineering student I would love to be able to follow this project from its infancy to first release.

Maybe a monthly newsletter could be released for it. Highlighting what took place. Or it could be every 3 months.

It could have a section for noob electronic engineering students and a section for advanced professionals. Universities could get involved and point there students to some material about it. eg the newsletter and other documentation.

These students could implement parts as there final year projects. Later when they are more experienced they may continue to contribute to it because they feel like its their baby.

:) Anyway just throwing ideas.

@Anonymous (IP:
by Jayclark on Mon 29th Nov 2004 09:36 UTC

Thanks for clearing that up. It makes a lot more sense to me now. I hope they do a wonderful job and accomplish want they set out to do.

You have forgotten the Noobs
by Ken Starks on Mon 29th Nov 2004 10:29 UTC

I remember the day vididly, when I had to recompile my kernel in order to get my ati POS card to work under Linux. I believe it was 3 re-installs because I couldnt replace my saved config file. If this card works "out of the box", it is going to help Linux be a player on the desktop field. If you still have to jump through all the hoops to get it installed...screw it: my nvidia will do nicely thanks.

to anyone believing that this is going to have an big inpact
by zaphod on Mon 29th Nov 2004 11:02 UTC

...look at the amount of available opensource cpus and other hardware:

so, anyone who uses an oss cpu from a little, rebellious cpu vendor, please raise your hands (btw amd is not the little sympathic alternative, but a multibillion megacorporation)...
anyone? ... see, thats the big impact another piece of oss graphics hardware is going to have.

by Treza on Mon 29th Nov 2004 11:59 UTC

Maybe most people here don't realize what a FPGA based design gives : Of course the performance levels cannot reach a custom made ASIC but a reconfigurable FPGA permits to enhance the design by uploading a new configuration, after you've bought the card. Reconfigurable HW permits as well to adapt the design to the software. One can imagine a Matrox G400 compatible mode activated by downloading the proper configuration ( So that board would work with Windows ... ). You could also expermient more easily alternate rendering techniques. For example, create a design geared only on 2D, saving FPGA area by removing 3D parts, which would do HW text rendering, for example. Others would prefer video codecs. One could imagine on-the-fly reconfiguration ( say, when you launch your video player )
A FPGA based graphics card could give amazing experiments possibilities.

RE: Multi-purpose.
by AndyFong on Mon 29th Nov 2004 17:07 UTC

For a $100, this card with a FPGA is a steal if you don't see it as a graphics card. Check out the any FPGA prototype board,
they cost way over $100. The cheapest I found is the Spartan III prototype board for $99.

How cool is it that you can actually learn VHDL/Verilog and have a card to play with which happens to have a lot of rams and a DVI interface on it.

So, for $100, I will buy it even without any code on the FPGA. For $200, I would maybe expect some good Graphics code on it.

Looking for Experts
by Savage Sailor on Mon 29th Nov 2004 19:03 UTC

Head over to the forums at the OGRE website, it's an open source 3D engine. Lots of fat 3D brains lurking, possibly with spare time and spare expertise.

by Wrawrat on Mon 29th Nov 2004 19:56 UTC

True, that would be very interesting but the boards would probably cost an arm and a leg unless they manage to get sponsored by a generous firm. I don't think we would see a decent board with a FPGA that is strong enough to do 3D rendering and with enough RAM for under 200$. And given that nVidia's mid-end offerings are around that price... I love openness but I am pragmatic.

re: anon et al
by Ryan on Tue 30th Nov 2004 06:30 UTC

These cards are already considered "open" and works on several different platforms without x86 legacy bios initialization.
This is false. G-series cards rely on the legacy BIOS for their power-on configuration information. Without the information in the BIOS, it would be impossible to know what card you had, what type of memory it had, etc so you would not be able to soft-boot it. The BIOS is not all initialization/int10 code.

Re: the lone oser
by Ryan on Tue 30th Nov 2004 06:32 UTC

I know Matrox have EMBM, thats why nVidia and ATI pay for that license,
Nobody but Matrox has DirectX EMBM. ATI has a similar extension for OpenGL but it does not describe a mechanism that is congruent to Matrox's.

re: BR
by Ryan on Tue 30th Nov 2004 06:33 UTC

They could purchase the PowerVR Kyro, and Tseng patents, and build from there.
What relevant patents did Tseng hold, and why would they have to buy them considering Tseng is long gone?

re: shurik
by Ryan on Tue 30th Nov 2004 06:35 UTC

Could you, for instance, use the NV_* extensions or are those under a proprietary license?
The extensions themselves are not proprietary, but implementing them in any sane way might infringe on patents.
3. XvMC ?
motion compensation is pointless given the power of todays CPUs. What is really necessary is hardware colorspace conversion and overlay.

re: zaphod
by Ryan on Tue 30th Nov 2004 06:42 UTC

see, thats the big impact another piece of oss graphics hardware is going to have.

RTFA. It's not an open source design. It's an open spec design.

Financing and Risk Sharing
by Bernhard on Tue 30th Nov 2004 13:55 UTC

Can't the guys from Xilinix just provide some vendor financing. Tweaking the payment terms is something that they could do without taking too much risk (getting a credit rating for TechSource shouldn't be that difficult). How much do 1000 Spartans cost anyway?
Also, wouldn't it be a great marketing opportunity for Xilinix, showing how powerfull their FPGAs are and that they can be used in mass consumer devices.

My two cents

RE:Will be a commercial failure
by Anonymous Bob on Tue 30th Nov 2004 23:10 UTC

Just because you an buy an nVidia TNT for $15 doesn't mean you're getting a better deal. If you actually read the specs, Techsource wants it to be 3D also. And no, it does not have to play the latest games fast to be a successful card. It would be a success if it worked decently with the OS and had decent graphics.

Good news
by Nice on Wed 1st Dec 2004 04:13 UTC

Very nice iniciative! Thank you Techsource for supporting Open Source community.

I'll buy this card, freedom is very very important to me.

Missing part
by Pootz on Wed 1st Dec 2004 11:51 UTC

There's something most of the people here is missing: opensource model of development has already showed us it's potential of doing things great and better (like OSes, network service, standards compliance, etc.). So let's give a try to this model in the hardware market. Of course the first (or second) implementation won't be the best 3d video card of the world, but soon it could get there. Remember: Linux was nothing but an academic "toy" 12 years ago, but now it beats OSes that is playing around for more than 20 years...