Author: Dann Corbit
Date: 11:58:02 11/11/98
Go up one level in this thread
On November 11, 1998 at 12:23:50, Will Singleton wrote: [snip] >I'm sure people are interested in such a test, but it sounds like a tremendous >amount of cpu time you're asking for. What do you estimate are the time >requirements? 6751 rows/30 computers = 225 rows per computer, each row analyzed at 12 minutes = 2700 minutes = 45 hours of computer time, normalized to a PII 300 Mhz (I have a large normalization table for other machines). About 75 hours for a Pentium Pro 200. I already have a SQL database model I am using for the Chess Analysis Project that can simply be copied and reused to handle all of the assignments. It would take *at least* 90 machines to do it right. 30 Crafty, 30 Hirarcs, and 30 Rebel 10. You run the program when you go to bed at night, and over the weekend. (This can obviously be flexed around a bit). In this way, the project would take only one weekend, if all goes extra-smoothly, and one week if everything possible goes wrong. If we got twice as many computers, it could be done very quickly. There are 4000 CCC members. If all participated, it would take: (6751 rows * 3 programs)/4000 machines = 5 rows per machine at 12 minutes = 1 hour. It sounds absurd, but the Mersenne Prime search project has over 4000 computers. Why can't chess get just as big an effort? Of course, some people have several machines. Of course, most people won't want to bother.
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.