Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Weird things with hash tables.

Author: Ulrich Tuerke

Date: 01:36:47 02/24/99

Go up one level in this thread



On February 23, 1999 at 23:42:41, James Robertson wrote:

>I have added what I think are hash tables. But, there are several things that
>have me worried: the biggest is that the pv will often change, e.g. in ply 6
>without hash tables one move will be chosen while in ply 6 with hash tables
>another will be chosen. Is this normal?

I think that you cannot exclude such effects, because the hash table modifies
the move ordering, and - at least in practice - the search result may depend on
move ordering. But I would be alarmed if it happened very often that my search
with hash table picks another move.

>From the starting position I am getting maybe 31% hash hits if the q nodes are
>not counted.
>I am getting almost 1 ply extra on average if I set the tables to 2 ^ 18
>entries. Are these normal?

Seems a somewhat large win to me, because you are talking about the starting
position. In early endgame / late middle game I would expect an extra ply. But
for the start position ... , but I am not sure.

Uli

>
>Thanks,
>James



This page took 0.01 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.