Author: Dann Corbit
Date: 14:39:24 10/19/04
Go up one level in this thread
On October 19, 2004 at 17:14:51, Robert Allgeuer wrote: >Following comments from people in the Winboard forum I have updated the recently >posted Zugzwang test suite; see the 30 positions at the end of this post. > >In order to see, whether results of this testsuite can say something about the >capability of an engine to deal with Zugzwang, I have compared 3 settings of a >beta version of Fruit: one with the verification search only in endgames (vseg, >which is default), one with the verification search always (vsalw) and one with >the verification search totally disabled (vsnvr). The results differ quite >significantly, as can be seen below. > >Time per position: 3 min >Athlon XP 2400+ >Rated time is the sum of solution times in seconds; 180 in case the position was >not solved. > > >Position Fruit X vseg Fruit X vsalw Fruit X vsnvr >ZPTS 1 0 0 0 >ZPTS 2 0 0 0 >ZPTS 3 0 0 0 >ZPTS 4 0 0 0 >ZPTS 5 0 0 0 >ZPTS 6 180 0 180 >ZPTS 7 3 0 0 >ZPTS 8 0 0 0 >ZPTS 9 23 20 38 >ZPTS 10 0 0 0 >ZPTS 11 180 0 180 >ZPTS 12 180 0 68 >ZPTS 13 0 0 0 >ZPTS 14 180 180 180 >ZPTS 15 180 0 180 >ZPTS 16 26 180 180 >ZPTS 17 180 7 180 >ZPTS 18 0 0 180 >ZPTS 19 180 180 180 >ZPTS 20 180 180 180 >ZPTS 21 180 80 180 >ZPTS 22 0 0 0 >ZPTS 23 180 0 180 >ZPTS 24 18 53 180 >ZPTS 25 180 11 180 >ZPTS 26 61 1 62 >ZPTS 27 180 2 180 >ZPTS 28 180 180 180 >ZPTS 29 0 0 0 >ZPTS 30 13 10 13 >Rated Time 2484 1084 2881 >Total Solved 17 25 15 It is clear that verified search always gives a significant boost in tactical strength for zugzwang positions. The other interesting question would be how do these settings affect game play?
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.