Author: Alessandro Scotti
Date: 03:10:14 08/16/04
Go up one level in this thread
On August 15, 2004 at 21:12:37, Andrew Platt wrote: >I just implemented this last week because of similar problems (I knew the >problems would exist when I implemented the hash table part but when >everything's broken in the engine it's somehow so easy to let others in). Just >make sure you debug it with some known positions because it's very easy to have >bugs with this implementation that turn out to be a lot harder to find than it >announcing mate in 80! This morning (before going to see Kasparov signing copies of his book) I added the fix and performed a couple of tests and it seems Mathieu was right on target: everything seems to be working much better now! :-) I always try to stress the engine with "ad-hoc" positions, but it's not easy to find good ones. It would be really useful to have something like the perft() test targeted to different parts of a chess engine. >BTW, I've never understood why you would implement just storing a bounded mate >in the table. Am I missing something with doing it this way that will bite me >later? I don't know, I am using MTD(f) and I try to store as much as I can in the hash table! Especially with the above fix, it doesn't seem to hurt at all...
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.