Author: Georg v. Zimmermann
Date: 02:54:06 08/25/02
Go up one level in this thread
Using the kind of extensions you describe should greatly improve your performance on GCP's WAC and other tactical testsets. If it doesnt, something is not only going sub-optimal but probably very wrong. You should do some statistics about the # of recapture-extensions, check extensions, forcing extensions etc. That should give you some ideas of what might be wrong. It seems like you extend a full ply on recapture ? That is too much. If you can't do fractional extensions you might want to consider limiting recapture extensions to the leafs or to the root or do something more scalable like if ((recapture) && (ply < currentDepth *2)) extensions += ONE_PLY; where ply is the depth from the root at this point in the search and currentDepth is the current iteration of iterative deepening. Best regards, Georg v. Zimmermann
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.