Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Sjeng UCI engine problem

Author: Gian-Carlo Pascutto

Date: 14:27:55 06/15/02

Go up one level in this thread


On June 15, 2002 at 17:14:01, Dieter Buerssner wrote:

>On June 15, 2002 at 12:54:10, Gian-Carlo Pascutto wrote:
>
>>Ok, I think I know what the problem is then. When Sjeng is
>>playing from tablebases it may sometimes move too soon, which
>>will confuse the UCI interface.
>
>How do you come to this conclusion? Don't all engines play very fast, when they
>are in a tablebase position? Or, do you think, that 100 us vs. 1 ms may make a
>difference? I cannot test this myself, but I have never heard of this problem
>for Yace. I think, Yace can move in some cases in 1 ms or faster, when in a
>table base pos. I would also think, that the UCI protocol would not have a
>problem with this scenario, at least in principle.

I'm basing it on what happened when I saw things go wrong.

During pondering, the UCI spec states that your program should
keep searching until told otherwhise. If there is a mate or
tablebase position, Sjeng may move prematurely, before it gets
an UCI command. This causes problems in Shredder.

--
GCP



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.