As part of our NVIDIA Tesla launch coverage took place at the product event in California, Beyond3D's Tim Murray and myself sat down with a trio of NVIDIA experts vested in CUDA and Tesla, in order to get a better idea of the products and where NVIDIA were taking them, both now and in the future. Those taped two-on-one sessions were transcribed so we could use them as article references for our Tesla and CUDA coverage. We present the transcripts here, unedited, for your reading pleasure. Please note that the conversations were slightly less than half an hour each, and that there's some slightly overlapping content in all three, so if you see Tim and I asking Andy, Ian and Dave the same rough questions from time to time, or referencing the other two-on-one sessions, it was all in the name of picture building and information gathering.
Some things might not make too much sense without the context of the presentations and conversations we had the day before, and the tense might get a little mixed in places, too, so beware. They weren't interviews in the traditional sense, then, but their content is worth letting everyone see to fill out the Tesla and CUDA picture in ways we might not be able to do in straight copy. Bear that in mind as you make your way through them. Those interviewed were: Andy Keane, the General Manager of the GPU Computing group; Ian Buck, ex Stanford under the legendary Pat Hanrahan and now Software Manager for CUDA; Dr. Dave Kirk, NVIDIA Chief Scientist.
In this interview we talk to Ian about the software side of CUDA, be that the engineering effort, future features and direction, the programming model and more. Ian's history with Brook and GPGPU (and more) at Stanford means he's one of a select group responsible for shaping the current parallel graphics processor programming landscape, in a non-graphics sense.