Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: new thoughts on verified null move

Author: Uri Blass

Date: 09:55:54 11/24/02

Go up one level in this thread


On November 24, 2002 at 12:38:48, Uri Blass wrote:
<snipped>
>>>BTW, when I get this working, I have a huge cluster of quad xeons.  I can
>>>certainly re-run all your test positions with Crafty, and let 'em run 32 at
>>>a time to depth 12, 13, and 14 to see what happens...  If you want to send me
>>>your positions, I should probably re-run the 9, 10 and 11 tests as well to get
>>>a baseline comparison between crafty and your program also...  but I have plenty
>>>of machine time to run deeper tests...
>>>
>>
>>Testing this idea at depths of +12 will shed more light on performance of this
>>algorithm in deeper searches. Because of its simple data structure Genesis is a
>>slow searcher (about 100k NPS on my 733 MHz) and so I couldn't afford a search
>>deeper than 11 plies. But it would be great if you run deeper tests with Crafty.
>
>I do not think that the fact that genesis is a slow searcher prevents search
>of more than 11 plies.
>
>Fritz can get usually at least 18 plies in some hours.
>If the problem is time then it is still possible that other people test genesis
>with search of 12,13 and 14 plies(I do not plan to do it but I guess that you
>may find other volunteers to test)
>
>Uri

Based on looking at your page it seems that the problem is that genesis with R=2
often needs something like 100,000,000 nodes to get depth 11.

I think that Fritz and maybe even Crafty are going to need less nodes to get to
the same depth but I did not check 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.