Joel Jaeggli wrote:
I should qualify what I was thinking a bit more.
On Thu, 5 Feb 2004, Joel Jaeggli wrote:
http://dri.sourceforge.net/doc/dri_driver_features.phtml
On Thu, 5 Feb 2004, J.B. Nicholson-Owens wrote:
<SNIP>
What you're doing is purposefully constructing a narrow situation and
rejecting any response that doesn't match your response. You do this to
make people believe that your way is the only way to play these 3D games. I
refuse to work within those silly boundaries. I demonstrated that for a
rather low fee one can purchase a great deal of functionality without giving
up software freedom.
Part of the problem we face at this point is the development model that got most of of the open source drm drivers created was capital and labor intensive and required a careful balancing act between the percieved needs of the of the graphics chipset vendors and the needs of the open source community. The capital that was expended before, will never be recovered (I'm speaking as a VA Linux and redhat shareholder here).
At this point, the graphics chipset manufacturers would be the logical supporters of this sort of effort, and by in large they were supportive when it was done on their terms, but the competetive edge they retain over each other (ati and nvidia) is so narrow, that their isn't room for the sort of friendly collaboration with a cummunity rather than individual group/single of developer(s) that would produce high-performance open source drivers that support that absolute latest chipsets. Personally I resent the fact that that the two fastest cards are I have in linux boxes are both ati r2xx chipset boards, a radeon 8500 and a 9000pro, but I appreciate the substantial efforts of tungsten graphics and the the weather channel that made even that possible.
I would just conclude by saying to ati, and any other hardware vendors that are listening. If there were and open-source driver for the r360 or r350 or nv35/38 I would buy one tomorrow.
I have a SiS 3D card for which an open-source driver exists for Linux. The open source driver even gives better performance than the SiS version but the hardware performance is pitiful. (BTW, I even had this card in a friend's oc'd AMD2800+ Windows machine and it still worked poorly)
I also have Nvidia card, closed-source driver available for no-cost from Nvidia. It works GREAT. Out with theSiS, in with Nvidia, set happiness=true;
What I think would be best is choice. If the chip and board manufacturers would document things so that open-source drivers could be written then I would have choice. I could then choose between the vendors proprietary driver or an open-source driver. If I had to pay for either of them that would just be input to the choice. I'd likely still go with the video driver that gives the best performance. For other components, where I may be inclined to twiddle, my choice may be different (I have no desire to debug video drivers).
Just my CAD$0.02
Mike
As an aside this paper is highly informative even if it is 6 years old:
http://dri.sourceforge.net/doc/vendor_relationships_paper.html
