[Punting to debian-x.] On Sun, Feb 27, 2005 at 01:23:58PM +0100, Francesco Poli wrote: > On Sun, 27 Feb 2005 12:57:41 +1100 Daniel Stone wrote: > > On Sat, Feb 26, 2005 at 07:06:11PM +0100, Francesco Poli wrote: > [...] > > > I can only cite a work-in-progress project: the Open Graphics > > > project. http://wiki.duskglow.com/index.php/Open-Graphics > > > > > > But if you know of a modern-day video card with > > > > > > * accelerated 3D capabilities > > > * published hardware specs > > > * Free BIOS and firmware > > > * Free DRI drivers > > > > > > that is already in stores waiting for my money, then *please* tell > > > us! > > > > You will not find a card anywhere with free BIOS and firmware. I say > > this because the Open Graphics card does not currently exist, > > I know: that is why I said "work-in-progress project"... As an aside -- do you have the source to your motherboard's BIOS? Firmware for all its chips? Firmware to your wireless card (on-chip or on-disk and loaded at initialisation)? Your network card? Your TV? > > and I > > will honestly be surprised if they get it up to any decent level of > > performance and features; doubly so if they make it worthwhile for > > $us200. But anyway, I digress. > > I hope a decent videocard will be out, but of course I may be wrong... > :( The problem with Open Graphics is that designing a chip to the level of today's common cards (even just going one step back to ATI's r3xx and nVidia's GeForce 5xxx) is next to impossible, which is why both of them have massive teams of incredibly skilled chip designers, and spend hojillions of dollars in R&D trying to beat the other. As I've said before, I will be amazed if the Open Graphics project even manages to get a card to r2xx-level (8500, 9[012]00, 9250) for its first release, even at $us200. And then by the time they get to their second iteration, r2xx will be not just two steps behind the curve -- ATI will have their r520 chipset out, and nVidia will have whatever they're planning next, and OG will still be left in the dust. This is no disrespect towards the people on the project, who are all, I'm sure, far more skilled than I am; it's just that if you are capable of designing a chipset to that level, you could walk into ATI or nVidia, name any number whatsoever, and have that amount of money arrive in your bank account at regular intervals. > > The closest you'll come will probably be an r2xx-class (RADEON 8500, > > 9[012]00), which has three out of four. > > AFAIK, ATI Radeon 9200 hardware specs are not completely published. > At least there are people who complain for lack of documentation. Interesting; the only things missing to my knowledge are HyperZ and Macrovision (which can be found with a small amount of guesswork anyway). Of course, there will always be things which are never documented anywhere -- ASIC bugs in particular revisions, bugs in vendor BIOSes, whatever. > Moreover, it's not a recent chipset: more modern Radeons are not (yet) > supported by DRI free drivers... Correct -- the entirety of r3xx and r4xx-series cards are missing support. The r300.sf.net project is working to get open-source DRI drivers on r3xx-class cards, however. > But we are going off-topic on debian-legal... > Maybe we should see what debian-x regulars think of this issue. Daniel, a debian-x regular
Attachment:
signature.asc
Description: Digital signature