Author: Albert Bertilsson
Date: 11:05:51 12/12/03
Go up one level in this thread
On December 12, 2003 at 13:47:20, Jaime Benito de Valle Ruiz wrote: >Does your program use hash tables in perft to store just nodes for a given >depth? > When doing perft if depth to search is larger than two, it looks in the hash table if has already been calculated. When a calculation for a certain depth is done, the result is written in the hash table if depth is larger than two. If the position is already used I overwrite only if nodes of the new calculation is larger than the existing item, that way I make sure not to overwrite positions that it has taken long time to calculate with fast calculated positions. I store the key, nodes and depth for each position. >Maybe for low depths using hash tables does not compensate for the extra >processor cicles, but maybe for higher depths is ok considering all the >transpositions that will occur in the tree. > >Ok, I'm probably re-descovering the gunpowder again and maybe everybody knows >this already... but the perft in my program does NOT use the hash tables >(although it update them with every "make"); it just verifies the correctness of >the move generator and the make/unmake functions. > >I guess that using more hash memory in Sharper reduces the analysis time so >much... otherwise it makes no sense. > >Just wondering, > > Jaime
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.