Author: Uri Blass
Date: 07:19:53 01/05/06
Go up one level in this thread
On January 05, 2006 at 08:50:47, Charles Roberson wrote: > > I beleieve the reason for this not being noticed 20 years earlier in chess, is >that in 1979 the search depths were very small and material with very little >positional knowledge was all you could put in a program. > > Today, we can search 12+ ply at tournament conditions. Thus, accurate >positional knowledge is more key than every before. > > The program I am working on has years of history performing better at blitz >than at long time controls. If it is relative to other programs then it proves nothing because it is possible that your program earns 100 elo from being 10 times faster when other programs earn 200 elo from the same speed improvement. If I were prunning too much, it seems that would >show up regardless of the time control. With bigger time controls, one can >search deeeper. It is possible that the pruning problem only happen when the depth is big enough. Did you test your program in tactical test suites to see if it solve more positions at long time control? I suggest that you try some test suites like ecmgcp and see if it solves more positions with more time. You may compare with other programs and if you need to multiply the time by 100 to increase the number of solutions from 100 to 150 when other programs can increase the number of solutions from 100 to 150 by multiplying the time by 10 then it means that your search is simply inferior search. Uri
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.