On Wed, Nov 06, 2002 at 01:12:52AM +0000, MJ Ray wrote:
Further to this, no the ATI drivers for the ATI Radeon 8500 (also the FireGL) are non-free
So sourceforge hosts non-free software at http://dri.sf.net/ and http://gatos.sf.net/ or are they software ones? If they're software ones, why do they need kernel support?
lots of the functionality is present in software only, ie they are not good for 3d work. When you want a fast 3d implementation you want to do it in hardware, thats why 3d cards were invented, the 3d texture mapping for example on the (i think) Radeon 8500 is done in software which is very very slow (if you are using 3d texture maps which most people will want) unless afaict you use the non-free drivers.
[...] and for the Radeon 9000 and 9700 there is no support afaict (either free or non-free) all of these being the fastest ATI cards at the moment.
They work via vesa for now, though, but not as accelerated as they could be. I hear talk about code in CVS for them, but evidently that's not end-user stuff now.
yes thats not a very good idea to buy a decent 3d card for linux to have it not work.
[...] were the first company to get all of their mainstream cards working in Linux with full functionality. [...]
For some value of "working" in most cases I've seen.
I have used the card sucessfully with the non-free drivers since the beta releases, I also support at least 100 workstations at work using the same drivers with different versions of the card, I have never seen the kernel have problems with them. Indeed the chipset in the motherboard of this computer is known to be very bad with the Nvidia drivers and supposed to have many problems but I have never had any problems... yet... ;)
I don't like it but there is *no* real choice for 3d hardware on linux. You either get drivers that are 80% there it seems or to get everything you have to use non free software :/
The other pertinent information is that of course there is much technology in the drivers which is why Nvidia do not want to open them to the world at large as they are afraid that ATI et al will steal their technology and incorporate it into rival products.
Of course, this would mean *more* natural selection of ideas, *faster* development of new features (to keep their advantage) and *better* performance for the users, but more *work* for the hardware manufacturers, rather than relying on copyright to get an above-market-value price for their work. We should not have any part of this perversion of copyright to holds back progress. That's the opposite of why it was devised.
I agree with this, just graphics card manufacturers sometimes use drivers to freshen their products 6-8 months after release to make them not look so bad in comparison the the competitors next big thing.
There is also a posssibility that there may be some non-free licensed code which nvidia use to make their cards and drivers work, of course this would still not stop them doing a free software release with non free bits missing, thus allowing the free software community to put the bits back in. Again this may give you similar results to the ATI drivers where lots of important features are missing :/
[...] What the free software community really needs in the way of support from the GFX manufacturing community would be the source of the windows drivers along with the specs of the hardware.
I'm not sure this would always help, as it might encourage people to repeat mistakes. Then again, experience does suggest otherwise.
I think that when you make a comparison between the speeds of these cards in Windows to Linux that there must be lots of interesting ideas locked up in there, at least porting these bits of the drivers to Linux could give you even greater performance than windows as the driver interface and HAL is much slimmer on Linux. Of course I could be wrong, but the real crown jewels for all manufacturers must lie in the windows drivers.
Apologies if any questions are stupid, but this is something I'm still learning about by asking.
Its something that I have learnt lots more about in the past few days, and of course the situation changes daily with new free software drivers being released. There is a mass of information out there and lots of it is contradictory and I havn't been able to read every source that I have found. I may spend a while looking at this again on the weekend and write up a paper with proper references in it etc. The only thing that does occur to me though is that currently 3d support on Linux is less than ideal. I will try to investigate further and let people know what I find.