On Thu, Jul 17, 2008 at 1:34 PM, terry mcintyre <[EMAIL PROTECTED]> wrote: > I do appreciate having some head-to-head competitions using > similar hardware, but I also want to see unlimited open class > competition, where people are able to run multiprocessors, and > learn to scale algorithms to big-n processors. We'll all have > 1000 processors on our desktops in a few years anyhow, might as > well iron out the problems now.
yes, but let's not get in a rat-race for access to the biggest computers. let's normalize hardware, while allowing multiprocessing and other architectural freedom! a good normalizing factor is transistor count, for which wattage provides a first-order approximation. THEREFORE, arrange a "Green Computer Go" tournament, in which the wattage used by each machine is well monitored. play a marathon to calculate rankings, then normalize based on each contestant's sum of game-relative joules[1]. the best "Elo / joules" ratio determines the winner. design secondary awards based on rank bands (kyu, dan, pro) and power classes (embedded, laptop, desktop, unlimited). Green Computer Go is pretty different. but it's very inclusive, well into the future. here's a separate idea that plays along well: virtualization could be used to supply scalable clusters on cloud computing systems (like Amazon's EC2, in particular), while guaranteeing accountable firewalls and joule estimates, and maintaining the privacy of a team's code. it's a hardware limitation, but it allows cash prizes in Computer Go tournaments with participants using clusters. it's also perfect for Green Computer Go. cheers, - Ryan Grant [1] for each game pairing, the game-relative joules is the total difference in energy used between opponents. it's positive for one player, and negative for the other, unless both use the same energy - then it's 0 for both. it's independent of win or loss. if your computing architecture requires higher wattage than the opponent, resign lost games quickly. joules used for pass-moves should be free. _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
