Search found 4916 matches

by Don
Thu May 01, 2008 5:50 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Scalability study for chess.
Replies: 13
Views: 5052

Re: Scalability study for chess.

Hi Bob, Yes, I know there are arguments for and against using node counts. However, when I use node counts for testing I am not concerned with the contribution of time control algorithms to the overall strength of the program. If I were testing the absolute strength of two different engines I would ...
by Don
Thu May 01, 2008 4:41 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Scalability study for chess.
Replies: 13
Views: 5052

Re: Scalability study for chess.

It's very odd that Tord recently posted about fixed node levels because this was a factor here. My testing method of choice has always been "fixed nodes" and I tested that way for years. However most programs do not seem to support this method of testing. Had Toga supported this I would have tested...
by Don
Thu May 01, 2008 4:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Scalability study for chess.
Replies: 13
Views: 5052

Re: Scalability study for chess.

From a casual glance at the game scores, I don't see an obvious difference in thinking time between the strong and weak versions of each program but I can average these up (per game) at some point and we can get a good estimate of how much more time one version uses per game than another. There pro...
by Don
Thu May 01, 2008 4:27 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Scalability study for chess.
Replies: 13
Views: 5052

Re: Scalability study for chess.

Will a program with a very good evaluation function improve more with depth than one with a poor evaluation function? I'm doing a scalability study for chess to help understand this. It's similar to other studies that have been done in the past which determines how much a chess program improves wit...
by Don
Thu May 01, 2008 4:11 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Scalability study for chess.
Replies: 13
Views: 5052

Re: Scalability study for chess.

It's very odd that Tord recently posted about fixed node levels because this was a factor here. My testing method of choice has always been "fixed nodes" and I tested that way for years. However most programs do not seem to support this method of testing. Had Toga supported this I would have tested ...
by Don
Thu May 01, 2008 2:42 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Scalability study for chess.
Replies: 13
Views: 5052

Scalability study for chess.

Will a program with a very good evaluation function improve more with depth than one with a poor evaluation function? I'm doing a scalability study for chess to help understand this. It's similar to other studies that have been done in the past which determines how much a chess program improves with...