Does anyone here know how to program the numerical array processors (shading engines, etc) on high-end graphics cards? nVidia provides a "zero-cost-but-not-open" language, CUDA, which can be used to program these teraflop cards, but the learning curve is steeper than I can manage alone.
I am looking at radio sidelobe patterns for my "open technology" server sky project, and discovered that emitter swarms shaped like flattened geodesic spheres have some remarkable properties. You can see a cheesy demo here: http://server-sky.com/gsr02-link Just downloading and displaying the flash video will max your CPU. It took more than 4 hours to compute, and a full scale map ( 1000x1000 km ) with a full size array ( 30,000 thinsats ) would take 6 weeks per frame on a 3GHz Pentium. The core of the loop is six lines of code, with 3 multiplies, 6 adds, a sin() and a cos(). So you can see why I would like to compute a tad bit faster ... I figure a game programming guru would be great for this. There are a few hundred members of the Portland Indie Game Squad: http://pigsquad.com/ Surprisingly, I don't recognize any names. Any "crossovers" in PIGS and PLUG on this list? Keith -- Keith Lofstrom [email protected] Voice (503)-520-1993 _______________________________________________ PLUG mailing list [email protected] http://lists.pdxlinux.org/mailman/listinfo/plug
