Author: Uri Blass
Date: 08:34:40 12/25/02
Go up one level in this thread
On December 25, 2002 at 10:09:27, Uri Blass wrote: >On December 25, 2002 at 10:03:13, Vladimir Medvedev wrote: > >>There are 2 main use-cases of "perft" command for user of your program: >> >>1. Measure [relative] speed of your move generator at given hardware >>2. Check your move generator and guarantee that it works without any errors. >> >>I think, that the second task is much more important for me as the user of >>Movei, than the first one (there are many other ways of measuring effective >>speed of the engine... and usually I'm interested not in pure move generator >>speed, but in overall speed of "movegen + search + extensions + pruning + >>hashing +..." combination). >> >>So, my advice is: >> >>do not specially change your move generator code just to increase it's speed in >>"perft" test. >>It should be the same (with all features and bugs) that your engine uses in real >>games. >> >>We don't play perft, we play chess :)) > >I think that it is possible to give the user to choose between the options. > >There are advantages for calculating perft faster. > >If a programmer want to debug his(her) move generator then it is faster to >compare with a program that do it faster. > >Uri I can add that I see nothing wrong with the perft game. It is the only game when movei can beat every free program that I know. 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.