AMD agrees to buy ATI for $5.4bn

BBC ^ | Monday, 24 July 2006

[rahbert]

I can't recall that pixel buffer rendering ever worked
right on the R200s even on the Windows platform. I remember
working on a bug similar to that when I worked there in 2002/2003.

It was on an example program written by a talented
Swede, "Humus". The R200s (esp 8500) also had problems
with games such as "IL2 - Forgotten Battles". Pipeline
stalls when executing a display list. I suspect that
the problem is more than just software. R300s never had
the pipeline stall problem with that game.

If you could port Humus' demos to X/glx you might see
a patterm.

[mazack]

For what it's worth, there was in fact a time that I do recall pbuffers working properly. I think this was near the first official/public release when ATI added third party support and decided to make their FireGL drivers work with normal Radeons.

But besides that point, I feel that a lot of what's "supported" by the hardware doesn't really find its way into the driver. Instead, it falls back to a software implementation ... probably because it's easier for them to do that.

I think that the whole ATI driver situation in linux is a mess. I bought an ATI card in the first place because I wanted to use open source drivers. However, I became frustrated by the incorrect lighting implementation in the DRI driver, so I switched to ATI's proprietary drivers. Some bugs disappeared, and others emerged.

I have also heard those developing on windows (notably John Carmack of id Software) complain about how the hardware was being under used by the driver. I remember reading in his .plan about developing the Doom 3 engine and testing beta hardware from ATI. He seemed pretty impressed by the features and capability of the hardware, but he thought that the drivers were just terrible. Actually, if I recall correctly, he also helped them patch bugs in the drivers to make things work properly. He also provided code for the UtahGLX project on XFree86 (way back when) to get Quake 3 to work.

And yeah, I have also heard about this sort of half broken hardware issue where the hardware theoretically supports a certain feature, but in reality, it will immediately hardlock the card upon accessing the register. This isn't specific to ATI though. Take a look at some of the card status pages over at http://dri.freedesktop.org ... it seems that other vendors have this problem too.

But right now, drivers really aren't my main concern. What I'm most concerned about is SGI's bankruptcy and this rumor about them selling OpenGL to someone. Honestly, I can see Microsoft buying OpenGL just to destroy it and find some legal reason to prevent people from using it. Although, I heard a rumor that Google might decide to buy SGI. I really hope that this is the case. OpenGL is great, and I don't want it to go away anytime soon.