Author: Uri Blass
Date: 02:00:45 11/06/05
Go up one level in this thread
On November 06, 2005 at 03:56:49, Tord Romstad wrote: >On November 05, 2005 at 18:29:34, Reinhard Scharnagl wrote: >> >>So for what should FRC be announced, if the engine already had signalled >>its FRC awareness to the GUI? An FRC enable engine always could play FRC, >>there is NO NEED to SWITCH its mode, like the Shredder GUI does, making >>the handling of Chess960 unnecessaryly amd incompatibly complicated. > >I agree with Matthias. That the GUI does not tell the engine that it is >playing FRC is very unfortunate. Technically speaking, you are of course >correct that an FRC engine can play normal chess, without an explicit mode >shift. However, many programmers *want* to do things differently in FRC >compared to normal chess. For instance, I have some rather expensive >evaluation terms which are disabled in normal chess (because they are >irrelevant in that case). Programmers who don't want to do things >differently in FRC can simply ignore the mode switch. I don't see any >problem here. > >Gothmog, my old engine, was written before the official FRC extensions >to the UCI protocol was introduced. I had to use the Arena specific >extensions instead. Because there is no explicit mode switch in >Arena, I had to produce two different executables; one for normal chess >and one for FRC. That's what I call unnecessary and complicated. > >The situation is not very different from analysis mode compared to play >mode. Some programmers, especially those with an asymmetric evaluation, >want their program to behave differently in analysis mode. Therefore, >we have the UCI_AnalysisMode (or whatever it is called) UCI option. >Those (like me) who do exactly the same in analysis mode as in play >mode just ignore this option. You don't think you will find anyone >who complains about it existence. > >Tord I do not understand how you do the same in playing mode and analysis mode. In playing mode if I find forced mate or get the maximal depth I play regardless of time control. In analysis mode I need to wait passively to another command in this case so there is a clear difference that is not based on asymmetric evaluation. time management is not something to check when you are in analysis mode. When you are in analysis mode and fail low you can solve the fail low and search later the rest of the moves. In playing mode it may be better in some cases not to solve the fail low and search the rest of the moves because if you try to solve the fail low first you will probably not have time to find a better move. I did not implement this idea in movei but I think that it may be logical to search different in different time controls. Uri
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.