Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hash hit != Hash full

Author: Andreas Herrmann

Date: 09:38:56 06/19/04

Go up one level in this thread


On June 19, 2004 at 12:23:15, Gian-Carlo Pascutto wrote:

>On June 19, 2004 at 12:19:02, Robert Allgeuer wrote:
>
>>Yes, I confused the two. But how would an engine measure how full its hash
>>tables are; looks difficult to me? Furthermore this would mean that in most
>>engines - as they do not clear hashtables after each move - this value would
>>have to approach 100% during the course of a game, which I did not observe
>>for a single engine so far.
>
>Positions pertaining to a previous search are not counted. i.e. For the
>purpose of this statistic you can consider the engine to *do* clear the
>hashtable after each move.
>
>At least this is how it works in Sjeng, but I suspect most engines work
>like this.
>
>--
>GCP

Do you count only the fill rate from one main hash table - or from all hash
tables?
Do you subtract the hash entrys, which are overwritten in the current search?

I ask this, because there is no clear definition inside the UCI protocol. And
that's the reason, why i think, you can't compare the fill rates between
different engines, if you don't know what a special engine author is sending
there to the GUI.

Andreas







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.