Using GITG to get the right features into ATI's GPUs
Are GITG developers still able to get the hardware features they want into future chips? I remember a comment from you on two-sided stencil being a direct result of developers asking for it for future engines. Is that still the case?
[Richard]: At ATI we have a process which guarantees that we listen to the key developers on a regular basis so that they can influence chip design. We go out to the top fifteen or twenty games developers and make sure that we listen to them before we explain what we plan to do. We do this on a regular basis so that we can make sure that our hardware always fits with what they need and with what they want to target.
You say that listening to the top fifteen or twenty game developers is enough for you to get an idea on future hardware direction, providing the features they'll need. Does that small sample of devs encompass enough general aspects of the hardware that you won't miss out on a possible future feature that a really popular game might need, from an upstart developer you haven't been to see?
[Richard]: We try to be inclusive, in the sense that we are always keen on receiving people's input. Many successful games developers have come from very small enterprises where individuals started small companies and created surprises for all the industry. For that reason we want to hear what everyone has to say.
But it's also necessary to keep that dialog confined to those most likely to have highly constructive and inventive input. We can't hold meetings of 500 or 1000 developers and get them all to fill in forms on what the hardware should be. Design by committee doesn't work well.
So, in the end we take a pool of talented, bright and demanding people and we ask them. We also review who is 'in' and who is 'out' on a very frequent basis.
Are ATI confident GITG is a process that contributes to ATI being able to keep on top of the 3D hardware game? Resting on laurels is something that bites you on the arse quickly in the GPU world, so I'm guessing that the developer-focussed GITG activities are something ATI actively rely on to create good hardware.
[Richard]: Resting on one's laurels is equivalent to suicide. In an industry like this you are quickly over-taken if you don't try to set an aggressive pace.
That's exactly what happened to NVIDIA with the NV30. They totally underestimated how aggressive ATI planned to be - and as a result their market share fell.
We take input from PC games developers, publishers, Microsoft, the OpenGL ARB, our hardware partners, the mobile market, console developers etc.
There's a saying that "There's no-one on this planet that you can't learn from". You just need to figure out who are the best teachers - and for that you gave to cast the nets far and wide.
