Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Gambit Fruit 1.0 Beta 4b is out now

Author: Dann Corbit

Date: 15:03:02 12/02/05

Go up one level in this thread


On December 02, 2005 at 12:39:19, Uri Blass wrote:

>On December 02, 2005 at 12:30:17, F. Huber wrote:
>
>>On December 02, 2005 at 12:13:46, Joachim Rang wrote:
>>
>>>
>>>>
>>>>But what´s a bit strange: I also don´t get any mate scores but simply an
>>>>evaluation about 40..60 !?
>>>>
>>>>Regards,
>>>>Franz.
>>>
>>>Bitbases just score win or draw or lost no distance to mate. The only way to
>>>disply this in a pv is a very high + score.
>>>
>>>Joachim
>>
>>Thanks for the info, Joachim -
>>I´ve never delt with bitbases, so I didn´t know this.
>>
>>As there already exist tablebases since a long time, there must be any
>>advantages for those bitbases!?
>>I´d guess faster access times or smaller size?
>>
>>Regards,
>>Franz.
>
>Of course faster access times and smaller size.
>
>It is not important to know the exact distance to mate when you are winning
>unless you have problems to win won positions but the main advantage of
>tablebases is not in winning won positions but in transition to won positions
>and programs can use tablebases only at the root and use only bitbases in the
>search when they are not in tablebase position.

The most important thing is that bitbases generally fit into memory.

The time to access a memory object is only about 200ns even with a cache miss.
The time to access a object on disk is thousands or millions of times slower
than that.



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.