Author: Michel Langeveld
Date: 03:27:53 03/10/04
Go up one level in this thread
>This could indeed be the reason, but I observed very different behaviour for my >program when having different sized TT's. The behaviour of the version that >stores the NULL-move results is very much influenced by the size of the TT, >while the version without the NULL-move results being stored is performing >better with increasing TT-size (32, 64 and 128Meg). >On the other hand, the NULL-move storing version is faster (and expands less >nodes) when using 64 MegaByte... (Could it be the magic number?) Maybe 64Mb is huge in the test you perform before the replacement algoritm has shows its strength.
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.