Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A CORRECTION!

Author: Uri Blass

Date: 09:05:56 12/30/01

Go up one level in this thread


On December 30, 2001 at 11:17:20, Severi Salminen wrote:

>>It would be much better to count QSearch nodes as node generated BY QSearch.
>>That means that the roots of the QSearches should not be counted.
>>
>>You want to evaluate the efficiency of your QSearch, so you need to count the
>>nodes GENERATED BY IT.
>>
>>The root of any QSearch has not been generated by QSearch, and so should not
>>been counted in your measurement of QSearch efficiency.
>
>Yes, let's do it this way then. Let all horizon nodes belong to "normal" nodes
>and all nodes _generated_ by qsearch to qnodes. Thank you. And please double
>check that the counting is working properly...
>
>Severi

Here are the result of my stupid engine
no hash tables
no null move
even no fail high and similiar search techniques to tscp when the only
difference is order of captures when in some cases I evaluate bad captures and
do not search them first(for example after 1.e4 e5 2.Nf3 Nc6 I evaluate that
Nxe5 is a bad capture so if I am not in the qsearch Nxe5 is not the first move
to search).


first number is number of qsearch nodes when the second number is the total
number of nodes including qsearch nodes:

depth 1:130,168
depth 2:1197,1486
depth 3:1408,3567
depth 4:2232,12695
depth 5:13342,96512
depth 6:68222,530576
depth 7:636800,4352366
depth 8:3461039,26576592

I guess that depth 10 may take some hours to complete so I am not going to do
it.

Uri



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.