Author: Steffen Jakob
Date: 21:44:07 07/28/98
Go up one level in this thread
On July 29, 1998 at 00:38:08, Danniel Corbit wrote: >On July 28, 1998 at 23:43:08, Steffen Jakob wrote: > >>Hi all! >> >>Definitely it's a good idea to support multiprocessor machines like crafty is >>doing. But, most people don't have computers with >1 CPU. More often they have >>access (at least at night :-) to >1 computers. Therefore my question: did >>anybody make experiments with chess programs and PVM? >It would be pretty hard to manage things like hash tables across a net. If you Yes, prob. you cannot do the same things which you do on a multiprocessor machine because the communication is much slower. But I think that there are many interesting ways how you can use PVM. >had one gigabit/sec netcards, it might be doable. Mostly, that won't work so >well on a 10 Mbit/sec net or even 100. I think something you could easily do >and it would be fun would be to have a voting system. Have Crafty, Comet, >Fritz, Rebel, etc. each running on a different node. At completion of the time >interval, they send a message to a master node with analysis data included. >This master node will examine the calculations of each independent node and make >a final decision. This would also be a good place for AI, as a simple system >like that would be fairly easy to "learn" from, since there are very few >external inputs into the system. > >I bet it would lead to a few surprises also. Hehe, nice idea although PVM would be an overkill for this. Greetings, Steffen.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.