Author: Bob Durrett
Date: 13:59:12 03/14/04
Go up one level in this thread
On March 14, 2004 at 13:03:09, Michael Nolan wrote: >On March 14, 2004 at 12:46:12, Bob Durrett wrote: >>I just notice that you said 6-12 hours "per move." I have never done that and >>cannot imagine why any "user" would do that. A chess programmer conceivably >>might do it out of curiosity to see what would happen to his chess program in >>that extreme situation. >>Perhaps you intended to say that you were interested in having the program look >>at a single *****POSITION***** for 6-12 hours. That is a very different thing >>from looking at a single move. > >Yes, Bob, 6-12 hours for a given POSITION. With several or many chess-playing programs, the user is given the option of setting up the program to automatically examine a given position. The size and shape of the analysis tree can be controlled to some degree by the user prior to starting the analysis. [The user can also preset the amount of time the program will consider each move.] In contrast to the above, the chess-playing program can be put into infinite analysis mode, which is different from the above. In this mode, it will continue computing the computer's next move until the user tells it to stop and display the result. In this case, the user will be given the move preferred by the program and may, or may not, be given a line following that move. This line is cryptically referred to as "the PV" by knowledgable chess programmers. The naming of "the PV" is most unfortunate. "Principal Variation" seems to imply that the sequence of moves is the best that has been found by the program at the point of time when that PV was first displayed. Unfortunately, some programmers program their software to output a sequence of moves which is reconstructed from hash tables. This process can be affected by the user because the user can control the size of the hash tables. The sequence of moves displayed may or may not have anything to do with the best variation and hence it is seriously mis-named. In my mind, at least, there remains much mystery about how data gets into the hash table(s), which data stays there, how long the data remains, under what conditions the data is erased or replaced, and similar questions. Unfortunately, every chess programmer creates and uses hash tables differently so that there is no standard way hash tables are used or anything about the data common from one program to the next. Anything that can correctly be said about how one program uses hash tables can be completely wrong when applied to another program. Disclaimer: All of the above "brilliant words of wisdom" are merely my understanding of what others have said here at CCC in the past. Bob D.
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.