29,90 €
Problem: I have often noticed something remarkable happening during the analysis of variations. It is apparently almost irrelevant which engine I am using (e.g. Komodo, the UCI version of Shredder 13 or others). I have also tried it alternatively when using Nalimov-Tablebases or Syzygy-Tablebases.
The most notable case is when in endgame positions (often close to the "Tablebase range") the engines show good results; but later in the variation we reach a position from which, for example, the same position can be reached again by playing Kd5, and all the engines show only this move in the Kd5 variation and the variation stops at this point. All that would not be so bad since if I play this move normal calculation continues. What is much worse is that two or three moves before possibly reaching the original position this is obviously wrongly evaluated or probably simply no longer appears. I had some examples in which a definite draw (+/-0.00) was indicated, although the original position can be brought about by force, a position which was previously evaluated as a clear win. Since it appears not to depend on which engine is in use, the reason can only actually be connected with ChessBase or general engine weaknesses (perhaps linked to Tablebases) or my computer. I have, e.g. a hybrid hard disk with 8 GB SSD. It is probably also nothing to do with hyperthreading, I have also made use of only one core. I would be glad if you have some idea, it is perhaps a known problem.
Explanation: Many engines evaluate a position repeated twice as a draw and thus stop calculating. As a rule anything else would appear to them as a waste of time, since computers “think” differently from people. We are well aware of this and try to avoid the breaks in analysis which result from it. Also the authors of engines try to avoid the most egregious blunders, but it does not always work out. The advantages of this implementation are much greater than these obvious disadvantages.
When analysing with computers, this aspect must always be taken into account and borne in mind. In many positions Tablebases are practically useless since there is too much material on the board.
Back to List
to send a comment please fill the form:
Name *
Email Address *
Subject
Feedback *
