Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Yace tablebase / bitbase test continued (long post)

Author: Heiner Marxen

Date: 09:25:06 07/22/03

Go up one level in this thread


On July 21, 2003 at 16:09:50, Robert Allgeuer wrote:

[...]
>As a conclusion it appears that tablebases and bitbases are no real advantage
>(which was also the result of previous experiments), although it is still not
>clear - at least for me - why this is so. Any ideas?

With respect to table bases the obvious drawback is the large cost of disk IO,
which costs nodes or even depth.  While a EGTB hit saves searching a subtree,
an equivalent result may often be achieved by a shallow search, which is
significantly faster than the probing.

For bitbases (which are much smaller) this effect should be much smaller,
but then, the benefits of bitbases are also smaller.

Obviously, probing EGTBs is a tradeoff.  Your results show, that the current
method for probing results in no measurable net gain.  Others have made
similar observations (me included).

I think, that a clever asynchroneous IO schema will be necessary to overcome
this.  Kind of SMP with more threads than CPUs... (wild guessing)

>Robert

Cheers,
Heiner



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.