Author: Bruce Moreland
Date: 10:42:59 09/08/01
Go up one level in this thread
On September 08, 2001 at 08:40:01, Kim Roper Jensen wrote: >Hi > >Just a question, when running test suites how much of a line should a engine see >before I can say it has solved it ???? > >I understand that it isnt enough that it sees/guesses the first move, but what >is your opinion ????? > >With regards When doing a large suite, solved is solved, because if you go through and look at every solution, you are going to go insane. With a smaller suite, if you solve a tactical problem in 1 second that used to take you a minute, often it's the problem's fault. The best tactical problems, if you are trying to test straight tactical vision, are tests where the program won't zero in on the winning move until it sees a big score jump, and where there aren't other positions that result in a similar big score jump at any point. But these might not be the positions that indicate that your program has gotten better. bruce
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.