Author: Sune Fischer
Date: 07:00:04 01/01/04
Go up one level in this thread
On January 01, 2004 at 09:44:04, Uri Blass wrote: >On January 01, 2004 at 08:37:03, Sune Fischer wrote: > >>On January 01, 2004 at 02:48:22, Russell Reagan wrote: >> >>>"Fast perft" is when someone calculates perft, but they use some tricks to >>>speedup the calculation, such as using hash tables to store previously computed >>>results (just like how the transposition table works). "Fast perft" has a >>>purpose, but not for this discussion. It would be useful if someone is >>>calculating perft to a high depth, and they wanted a fast perft calculator. >>> >>>"Fast perft" doesn't really compare the speed of the movegen/make/undo routines, >>>and it doesn't help you to debug your search. >> >>Of course it helps you to debug, think about it. >>A hash hit just saves you a perft calculation of the same position to the same >>depth. >>You've already counted this number, either it is correct or not but there is no >>need to perform exactly the same task again. > >There is a need to do it to verify that the number is the same. > >It is possible that you have a bug that cause the program to be right in the >first time but wrong in the second time because of some varaible that was >changed. It is also possible that if you first do a perft 4 then a perft 5 then a perft 4, that the last perft 4 will result in a different node count. You should check that to be _really_ sure....:) And if you want to be really _really_ sure, you should do a perft 4 followed by a perft 5 then a perft 4 (confirm now the two perft 4's) then you do perft 3 and then again a perft 4 (which is now compared to the first two perft 4's). But to be really really _really_ sure you should first a perft 4 then.... :) -S. >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.