Greetings,

I am thinking about building a multi-headed machine for multi player "network" gaming.

I need to put 4-6 OpenGL compatable cards in the machine to make this happen.

I have seen much documentation of ati and nvidia cards, but I have no interest in subsidizing companies who don't release specs.

My current laptop has an intel card onboard which uses the free (as in speech) i810 driver and has good enough OpenGL support to allow good frame rate play of Quake3 engine based games, which is most of the well polished net games under linux.

Can anybody point me to pci or pci-e i810+ based cards, or another fast-ish non-blob OpenGL card which I can use for this project?

Matrox makes high quality cards, and they do multi-head quite well. I think they must be linux-friendly, as their drivers always seem to be the first and best supported. I couldn't tell you anything about pricing though.

According to the wikipedia page on matrox all their linux drivers are binary only or binary-blob dependent.

If I simply have to build a cluster of laptops with on-board 945GM chips that would work.

My criteria is simple, in five years when the vid card is no longer supported by the manufacturer I still want to be able to run it on a new install of linux/BSD, only fully open driver make that possible.

does anybody make an add-on board based on the 945GM chip set?

It appears that you are correct. However their recent cards have very good open-source drivers. So I guess if you simply didn't want to install proprietary drivers the Matrox would work. If however, you want to stand on principle and refuse to buy from someone who won't distribute open drivers, I suspect you'll have trouble finding good quality, multi-head cards. However, I wish you the best of luck, and if you do find a good one, please report back. I know I would be very interested in one.

P.S.
I don't know if it makes any difference to you, but I think I remember hearing that a representative of NVidia said that they would release open source drivers, except that other companies' IP was involved. So, assuming that I'm remembering this correctly and that NVidia isn't just telling a white lie, then they would, but their hands are legally tied.

I grabbed the source code off the Matrox site, unfortunately it is dependent on a sizable closed-source library, the documentation of which says:

like many other linux-able vid-cards this driver is only part "open", but none of it is free as in speech by GNU or BSD standards (which is relevant to keeping legacy drivers running).

Still looking for any leads on non-binary dependent OpenGL vid cards.

I know that Matrox has their own "open-source" driver, but I was under the impression (and I may be completely wrong) that there was a separate, really free driver worked on by folks not affiliated with Matrox.

Okay, after a little more looking around and considering Patrick's observation on redundant free/closed drivers I found http://www.free3d.org/. While it seems that nobody making decently powerful video cards is playing nice lots of hackers still go to the trouble to reverse engineer and build free drivers for them none the less (I wouldn't, but that's me).

So unless anybody has any more leads I guess the answer is "No...but". I can hand AMD/ATI some cash and use free drivers which they made difficult to develop.

From a practical perspective I can run decent video cards on free(therefor not fundamentally broken) drivers.
From an ethical perspective I really don't want to had the money to AMD/ATI for chips they basically don't support.
I suppose I'll probably end up splitting the difference and buying used ATI X800 cards on ebay.

Indeed, buying used does seem to be the solution to your dilemma. You avoid financially supporting closed, proprietary drivers, yet still get pretty good performance. I may do this myself in the future.

I only wish I had an option of somebody to support. Acting by withholding support does not feel like a very proactive solution to what is in essence a much broader problem. Not an easy itch to scratch though.

I would consider the open-graphics project, if it were not so pre-alpha (as in no usable cards yet).