Trends in GPUs
HC: Do you see NVIDIA moving to the strategy AMD used with the launch of the 4800 series and focusing on developing a mid-range GPU with the option of doubling up?
JP: First off let me explain that Nvidia is a name, not an acronym, so it shouldn't be all caps, unless they are sending you a cheque to do that. I think Nvidia will have no choice. ATI has the right idea and it looks like it's paying off for them, subject to their ability to market it (they actually lost market share to Nvidia in Q4 - go figure.)
It's a three-prong step: 1- reduce chip size by reducing some performance, 2- reduce memory bus width, and 3- jump to a very fast memory (i.e., GDDR5). ATI has one more trick up their sleeves which may be exposed in Q2.
"ATI has one more trick up their sleeves which may be exposed in Q2"
HC: Do you see NVIDIA and AMD shifting their focus more towards mobile GPUs and/or cheaper GPUs and how do you think this will manifest itself?
JP: Mobile is certainly where the growth is, and it comes with some serious challenges due to form factor thermals and power supplies, so it's almost a completely separate design. Cheaper GPUs are either derivatives of the top end design, and/or yield-managed binn'ed parts.
However, as Peddie's first law states: In computer graphics too much is not enough. So there will be no let up on the demand for better, more powerful graphics GPUs and due to the thermals and power supplies they will have to be used on (large) AIBs. But the volume will certainly be in mobile and low cost.
