Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: significant math

Author: Robert Hyatt

Date: 13:30:28 11/19/02

Go up one level in this thread


On November 19, 2002 at 16:27:31, Russell Reagan wrote:

>On November 19, 2002 at 16:18:48, Gian-Carlo Pascutto wrote:
>
>>On November 19, 2002 at 16:08:58, Russell Reagan wrote:
>>
>>>On November 19, 2002 at 15:37:24, Gian-Carlo Pascutto wrote:
>>>
>>>>Please show EVIDENCE.
>>>
>>>Crafty gets 302,000 nps on my machine. Yace gets 267,000 nps on my machine. That
>>>looks like bitboards are more than breaking even to me.
>>
>>Now try a 'perft' run...
>
>Why are we resorting to instances which will never be of practical use in a game
>of chess? Maybe after we compare perft's we can compare something equally useful
>like who has the larger opening book and base the "bitboards vs. non-bitboards"
>discussion on that, since that is equally unrelated to this discussion.
>
>If you lose a game to Crafty, are you going to say, "But my perft times were
>better!"?


It is a religious argument.  If bitmaps are worse in move generation, then they
must be
worse everywhere.  Even when there is ample evidence that they lose here and
gain there,
and that overall on 32 bit architectures they appear to be a break-even affair
at present.  As
your NPS showed.  Of course if you are going to rely on perft, which is move
generation
speeds (and which is certainly not optimized in the perft I wrote as it was just
for testing
and not for comparison to others) then you are relying on the speed that only
accounts for
10% of my execution time.  I find it a good practice to spend time working on
the parts
that take 50% and up, such as evaluation.  Perhaps we need a "perft" option that
includes
an evaluation for every position generated...




This page took 0 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.