Radeon DRI for FireGL T2

Mike A. Harris mharris at www.linux.org.uk
Tue Aug 31 12:00:37 PDT 2004

Vladimir Dergachev wrote:
> On Tue, 31 Aug 2004, Mike A. Harris wrote:
>> Hamie wrote:
>>> Question is, whats the status of support for the 3D acceleration of 
>>> the FireGL-T2... 
>> The status of 3D acceleration for R300 and newer chipsets, is that 
>> there isn't any.  There is nobody working on them, and there are 
>> currently no specifications available for the 3D part of the chip.
> Small correction: I am trying to figure out how to use R300 3d pipeline
> (http://volodya-project.sf.net/R300.php) but the progress has been slow 
> so far.

Yeah, it's good that you've taken an interest in this, and started it 
off.  The core problem I think though, which I attempted to state in the 
first message, is just how complicated the problem can be to solve, and 
just how many man hours may be involved.  Without funding, and without 
knowledge of the hardware from specs, it's more or less a very long 
windy road, and only then if someone has the time to dedicate to doing 
it more or less full time.  ;o/

It'd be great to see something come out of the above work, but does 
anyone have that much time to dedicate to it?  I can see many people 
having vim and vigour to start trying to figure out such chips operation 
and/or reverse engineer them, but personally, I think the further 
someone delves into the problem, the more complicated they realize the 
problem is, and how much time it would take to make even small progress, 
and they lose interest in doing it, or just don't have the time to 
dedicate to it due to real job, or other matters in life.

So, while I applaud anyone and everyone who tries to do such an uphill 
task, I still think it near impossible that we'll see a fully 
functioning OSS R300 driver without help from ATI (microcode, PRG, DDK), 
and funding from someone to see it through completion within a timeframe 
that the hardware is still considered relevant.  The R4xx cards are 
current now however, and by the time such drivers might be completed 
under proper funding though and get the kinks worked out, a new 
generation of hardware is likely to be out, and the R300 work will be 2 
or more generations old.

I'd desperately like to be wrong, and to see an OSS R300 driver before 
then, but I just don't think the manpower is available in the community 
to dedicate to reverse engineering the hardware and doing the necessary 
work in a reasonable timeframe.  People who have the level of skill who 
could pull it off, would probably be hired by a video vendor or other 
hardware vendor before they completed such a complex task, and would 
likely not have the time to dedicate to completing the work.

Again though, I hope I'm wrong, and look forward to being so.  ;o)

More information about the xorg mailing list