Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: fine #70

Author: Tim Foden

Date: 14:09:16 08/12/04

Go up one level in this thread


On August 12, 2004 at 07:50:42, Jan K. wrote:

>Hi...if you take hashHit%=(number of hash table probes)/(number of cutoffs), how
>high should it be in this position? Does anyone have numbers for
>"replace-always" and "replace-if-same-depth-or-deeper" schemes? Do I guess right
>that in this position "replace-always" will do better than other schemes?
>Thanks.

Green Light gives these statistics.

Here are the relevant lines after running the fine 70 position for 60 seconds:

trans:  probes=40932103  hits=29939232 (73.14%)  draft=25800493 (63.03%)
tcuts:  exact=10354 (0.03%)  upper=20542982 (50.19%)  lower=2081772 (5.09%)

So, from 40932103 probes of the hash table, the position was found 73% of the
time, and was usable for cutting due to the depth stored in the table being OK
63% of the time.

It actually managed to make cuts 55% of the time (0.03% + 50.19% + 5.09%).

Cheers, Tim.



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.