Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Thanks! (and some additional discussion)

Author: Artem Pyatakov

Date: 09:00:00 08/17/01

Go up one level in this thread


Heiner,

Thank you VERY much for this very detailed and helpful post.

Your technique of multiple hash probes is very interesting - I might experiment
with this in my newbie engine and see what kind of results I get.

Now, you talk about the problem that I did consider right after posting my
message - namely that the number of nodes that HAD to be searched could
significantly differ from the number of nodes that will have to be searched
later.

One (somewhat) interesting thought that I had on the subject, would be that
instead of counting the number of nodes ACTUALLY searched with hash hits, we
could compute the number of nodes that would have to be searched without the
hash table!  To do this, every time we get a hash hit, we would ADD the number
of nodes that was recorded in the hash table to the total number of nodes that
we searched.

I think getting this number would serve as a better estimate, because DEPTH also
approximates the number of nodes that would have to be searched WITHOUT the hash
table.

But again, there are definitely problems with this approach, since it assumes
that we have have no hash table to estimate node counts, which is certainly not
true.

Thanks again!

Artem



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.