Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Let's talk about fraud.

Author: Robert Hyatt

Date: 07:59:09 05/03/04

Go up one level in this thread


On May 03, 2004 at 01:50:44, Gian-Carlo Pascutto wrote:

>On May 02, 2004 at 21:03:06, Robert Hyatt wrote:
>
>>But it is within the noise
>>range for non-deterministic behavior anyway...
>
>I showed that the difference was statistically
>significant, so that argument holds no water.
>
>--
>GCP


Fine, then hold _this_.

We both ran the positions Vincent claimed produced _no_ speedup whatsoever for
Crafty.  You got 2.8x overall on a quad 550 of mine you were using.  I got 3.0
on my quad 700.  I sent you _both_ the logs.  I then ran the same test with
null-move disabled (sel=0/0) and got a speedup of 3.1

If you want to claim that 3.1 over 2.8 or 3.0 is that significant, go ahead.  I
don't.  I found that turning null off increased the speedup by 3.33%.  If that
is significant to you, fine.  To me, based on Vincent's attempt to explain his
poor SMP search results by blaming null-move, it is _not_ significant.

Statistics or other hand-waving notwithstanding....

If you take your numbers it is a 10% change.  That is _still_ orders of
magnitude from what he needed to explain the difference between his search and
the DTS results...



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.