Author: Ren Wu
Date: 14:25:45 06/22/98
Go up one level in this thread
My way of doing things is always to let the machine making the decisions, as many as possible. So If I tell the machine when to search killer moves, I somehow broke this rule myself. So I just assign a bonus to the killers, and of course i have many other rules to assign bonus to move too, like history table. I sum all the bonus together, and search from the move with biggest bonus. With some 10s of experiments, i can find the reasonable good value for each different of bonus. And the program works better this way. A lot better in my case. For the bk test, position #5, here is the node count of my program 9. 2883436 10. 5814088 11. 16739070 12. 36562629 It is a little better than crafty did at this position, even though it is still worse than DarkThought. Note that i don't use swapoff at all at this moment. So i think the move ordering is works pretty well for me. Ren. (renw@iname.com)
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.