Author: Ricardo Gibert
Date: 01:27:05 07/26/02
Go up one level in this thread
On July 26, 2002 at 04:13:16, Russell Reagan wrote:
>On July 26, 2002 at 03:33:28, Ricardo Gibert wrote:
>
>>Information about ruffian(C) (Last disconnected Tue Jul 23 2002 11:49):
>> Standard 2723 337 110 106 553 2723 (23-Jul-2002)
>> 2: Hardware: Athlon 1.4Ghz
>>
>>As you can see, it is _not_ an Intel CPU, but rather an Athlon. There is
>>_nothing_ unusual here. It is about 300 rating points weaker than a commercial
>>program running on the same hardware.
>
>On ICC Junior runs on a dual 750MHz machine and is rated 2701, Shredder plays on
>a dual 1.26 GHz machine and has a rating of 2713, Fritz plays on a dual 1GHz
>machine and is rated 2768. Ruffian seems to be just as strong as any of the top
>commercial engines. I don't see why you say it's 300 points weaker.
>
>Russell
I just glanced at the top single cpu system blitz rating and saw this:
Information about ManvsMachine(C) (Last disconnected Wed Jul 24 2002 21:55):
rating [need] win loss draw total best
Blitz 3277 916 80 166 1162 3327 (15-Jun-2002)
1: --->>> CURRENT ENGINE: Yace 0.99.56 <<<---
2: --->>> MAN vs MACHINE RATING PROJECT <<<---
3: PURPOSE: TO DETERMINE BLITZ RATINGS OF VARIOUS WINBOARD ENGINES IN HUMAN
PLAY ON ICC.
4: HARDWARE: 1.0 GHz AMD Athlon, 768 MB RAM, Win 2000 Pro, cable modem
connection.
5: SOFTWARE: Winboard 4.2.6, opening book ON (if any), ponder mode ON,
hashtable = 128(or 96) MB(if user specified), 3/4-man EGTBs only (if
used).
6: METHOD: Each engine begins with a 2200 estimated blitz rating. Rated
blitz against registered & established ICC human players ONLY. Autocolor.
MAXIMUM OF 4 CONSECUTIVE GAMES versus the same opponent.
7: DATA: Games with no final result, <5 moves and adjourned games will be
discarded. A minimum of 300 games will be collected.
8: RESULTS: Game collections and statistics are available at
http://manvsmachine.blogspot.com
9: Questions/comments about the Man vs Machine project should be addressed
to
manvsmachine@winboard.info
And then I made what I thought was a conservative assumption. My mistake. Still,
I see nothing unusual with Ruffian. Even if the hardware were running at half
speed, I would only expect a difference of 50 rating points then if it were a
top commercial program. Instead it is much less than that. The speculation is
not justified.
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.