Author: Dann Corbit
Date: 15:22:06 04/21/05
Go up one level in this thread
On April 21, 2005 at 17:00:19, Dieter Buerssner wrote: >On April 21, 2005 at 15:37:02, Dann Corbit wrote: > >> Some of the things I particularly like are the excellent annotations >>at the end of a search using EGTB, the auto feature to play against itself, and >>this looks like a very good and valuable tool as well. > >>Perhaps you can publish a formal definition of your extensions and some of them >>might be added to the Winboard and UCI protocols. > >Thanks, for your kind words. What I implemeted as "cook finding mode" is already >possible with the UCI protocol. The UCI protocol has the "searchmoves" to >restrict the choices to search in the root position. An UCI GUI could easily >"drive" the engine, to only search the moves that are not after bm, and you have >the same functionality given by the cook command in Yace. > >Auto play is also possible, with UCI and WB. However, the GUI would need to >start two instances of the engine, which has some disadvantages. Most resources >will be needed twice, compared to an auto play mode inside a single engine. UCI >mode also has the "info string" which can easily support all the stats yace >shows in console mode. WB has a similar feature, that unfortunately at the >moment is practiacally of little use, because it will always pop up some >message, where you have to click or hit ENTER (from memory, I did not check it >at the moment. I think it is "telluser"). This could almost be considered a bug >of WB/Xboard (the GUI, not the protocol). The protocol could be enhanced here, >too. The functionality is there, in principle. Two separate engines will probably not share the hash table. The thing I like best about auto is to analyze at one hour per move. It is very often that a chess engine after a one hour think will get the pm right, but very early on the pv goes awry. So with 1 hour thinking and an auto command, I can get a very good sort of analysis. Because the hash table is used by both sides, there is no loss from move to move. There is no better way to do it that I can see.
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.