Author: Steven Edwards
Date: 03:45:44 07/06/05
Symbolic: Running BWTC After three minor bugfixes, Symbolic was tested against the entire 1,001 position suite BWTC. There were no crashes or hangs, just correct key moves for 822 of the 1,001 problems. Observations: 1. About 450 of the problems are MateIn1 or MateIn2; these are solved immediately with two and four node searches. Symbolic uses the ChessLisp intrinsics for MateIn1 and MateIn2 recognizers, so these problems did not activate the full cognitive search. 2. Of the remaining solved problems, not all solutions have an exact analysis because Symbolic's defensive move suggestors are rather limited at this point in development. 3. Most unsolved problems can very likely be easily solved mith minor adjustments to the progressive deepening. 4. Deactivating narration generation gives a twofold speed increase. 5. Rarely did an analysis take more than ten seconds; most were less than five seconds. Node counts per search ranged from two up to about two hundred.
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.