Author: Daniel Shawul
Date: 04:37:21 08/20/04
Go up one level in this thread
On August 20, 2004 at 03:50:21, Ross Boyd wrote: >On August 20, 2004 at 01:55:19, Daniel Shawul wrote: > >> >>i used to work my chess on a 1.8GHZ PC. >>On that PC DanChess(slower program) can score good point >>against some fast programs with high NPS. But now when i >>changed to a 2.8GHZ PC things are different. my prog has >>really a very hard time against those engines. My question is >>both programs are speeded up by the same amount,so why the difference? how much >>are the other programs benefited? usually i test at 40 moves in >>3 minutes.what is the equivalent of this on a 2.8ghz pc. > >My guess is too many extensions. Extensions are effective on slower PCs but can >cause tree explosions which then limit the overall search depth. You could try >applying certain extensions (eg. recaptures, pawnpushes to 7th, simplification >into pieceless endgames) only when remaining depth <= 3 ply - or something in other words , to be less tactical. other than those extensions i also do markoff extensions , which seemed to work for me but others found it bad. like >that. > >Next guess is to improve move ordering and quiescence pruning. Write a SEE >routine if you haven't already and use it to prune losing captures in the >qsearch. All these things drastically reduce the size of the tree... i don't think size of my tree is that large. it falls on the 2-3 range. Problem is i think the large attack table generator in eval which is taking more than 60% of eval time.i am not going to remove that because i am using it for lots of purposes right now. my choice is to take advantage of it for pruning based on eval which might reduce tree size. thanks daniel > >Hope this helps, > >Ross
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.