Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: To engine authors: Does "force" clear search data?

Author: Tim Mann

Date: 20:56:48 01/05/01

Go up one level in this thread


John sent me some email about this, and so I can explain where he is coming
from.

CM8000 has a feature that allows the user to pause the game and stop the clocks.
They would like this to also pause the engine's search (or pondering), but the
WinBoard protocol doesn't document a way to do that.

John's question was whether "force" causes the engine to clear its transposition
hash table.  If it doesn't, then if the GUI wants to pause the search, it could
send "force".  It would later send "go" (or "playother" if the engine is not on
move and supports the playother command) to resume the search.  If the hash
table was not cleared, then the search should quickly get back to roughly where
it was when paused, achieving the desired effect.

Gian-Carlo is of course correct in saying that the protocol doesn't specify what
should happen in this case, so some engines may clear the hash tables and some
may not.  (Also, of course, some simple engines don't even have hash tables.)

I'm thinking of adding another optional feature to the WinBoard protocol that
allows the engine to say "it's okay to use force to pause and resume a search".
Or maybe instead, I would add a new optional command "pause", which is the same
as "force" except that the engine tries to save its search state so that it can
resume if "go" or "playother" is sent without changing the position.  Not
clearing the hash table would be one good way to implement saving the search
state, though an engine author could choose to do it some other way.

Comments are welcome...  I am going to post this to the chess-engines mailing
list too.

	--Tim



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.