Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fruit 1.5 Hash Hit Percentage

Author: Andreas Herrmann

Date: 08:40:54 06/19/04

Go up one level in this thread


On June 19, 2004 at 05:56:46, Robert Allgeuer wrote:

>
>I am currently running a small test tournement with Fruit 1.5 against various
>engines. Arena reports for Fruit consistently a comparatively low hash hit rate
>of 5-7%. Certainly my time control for this test is very short (2m+1, Athlon
>XP2400+), but other engines report under the same conditions significantly
>higher hit percentages (Patzer 85%, Ruffian around 50%, Anmon 40-50% etc.).
>Any ideas why?
>
>Thanks
>Robert

At first you can't compare the values, because AFAIK there is no definition
inside the UCI protocol, what to output there. So only the author of an engine
knows, what he sends there to the GUI.
The engines are using different kinds of hash table implementations, some are
using only one hash table, but most are using different hash tables intern. So
what value from which hash table should be send to the GUI?  And most engines
don't clear the hash tables after each move, so the fill rate should increase
during the game. The best is, you ignore these hash filled values, or you ask
the author of a special engine, what he sends there to the GUI.

Andreas




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.