Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rebel 40/2 against IM/GM's all 20 games.

Author: Chris Carson

Date: 05:05:33 01/11/00

Go up one level in this thread


On January 10, 2000 at 22:10:27, Jeff Lischer wrote:

>On January 09, 2000 at 20:23:26, Stephen A. Boak wrote:
>
>>On January 09, 2000 at 14:28:59, Chris Carson wrote:
>>
>>>On January 09, 2000 at 14:19:43, Wybe Koopmans wrote:
>>>
>>>>On January 09, 2000 at 11:25:25, Chris Carson wrote:
>>>>
>>>>>Program	Hardware	Perf.	Opp	Games
>>>>>Rebel	AMD K6II-300	2427	2427	2
>>>>>Rebel	AMD K6III-400	2515	2398	3
>>>>>Rebel	AMD K6II-450	2595	2795	2
>>>>>Rebel	AMD K6III-600	2645	2528	3
>>>>>Rebel	Kryo K6III-600	2470	2542	5   *
>>>>>
>>>>>* - 1 game (Hoffman) lost due to HW problems,
>>>>>    PR 2552 without this game !!
>>>>>
>>>>>Rebel	PII-350		2432	2432	2
>>>>>Rebel	PII-400		2466	2466	1
>>>>>Rebel	PII-450		2185	2585	1
>>>>>Rebel	PIII-500	2866	2466	1
>>>>>
>>>>>Source: Rebel web page:  http://www.rebel.nl/edindex.htm
>>>>>
>>>>>Team Fritz, Ferrett, Shredder, Silkchess, Junior vs
>>>>>Team Lutz,  van Wely, Sokolov, Vaganian, Alterman
>>>>>     (2600) (2635)    (2620)   (2630)    (2625) =  2622
>>>>>
>>>>>Programs score 2.5/5 for 2622 TPR at WCCC 1999 !!
>>>>>
>>>>>Source:  WCCC 1999 web page: http://www.uni-paderborn.de/~wccc99/
>>>>>
>>>>>Performance calculated using web:  http://maskeret.com/mecca/
>>>>>
>>>>>Raw data:
>>>>>
>>>>>Player	FIDE	Rebel 		Hardware	TPR	Opp
>>>>>Sakalauskas	2450	0	AMD K6II-300	2427	2427
>>>>>Zapolskis	2399	1	AMD K6II-300	2427	2427
>>>>
>>>>Hardware was: PII 350 Mhz
>>>>
>>>>>Russek		2398	1	AMD K6III-400	2515	2398
>>>>>Russek		2398	0.5	AMD K6III-400	2515	2398
>>>>>Russek		2398	0.5	AMD K6III-400	2515	2398
>>>>>Anand		2795	0.5	AMD K6II-450	2595	2795
>>>>>Anand		2795	0	AMD K6II-450	2595	2795
>>>>>Rohde		2585	0.5	AMD K6III-600	2645	2528
>>>>>Ruzele		2515	0.5	AMD K6III-600	2645	2528
>>>>>Zagorskis	2486	1	AMD K6III-600	2645	2528
>>>>>Baburin	2593	0.5	Kryo K6III-600	2470	2542
>>>>>Sherbakov	2552	0	Kryo K6III-600	2470	2542
>>>>>Scherbakov	2541	1	Kryo K6III-600	2470	2542
>>>>>Sorin		2524	0.5	Kryo K6III-600	2470	2542
>>>>>Hoffman        2501	0	Kryo K6III-600	2470	2542
>>>>>Krivonosov	2432	0.5	PII-350		2432	2432
>>>>>Krivonosov	2432	0.5	PII-350		2432	2432
>>>>
>>>>Hardware was: AMD K6II-300
>>>>
>>>>>Grabliauskas	2466	0.5	PII-400		2466	2466
>>>>>Rohde		2585	0	PII-450		2185	2585
>>>>>Grabliauskas	2466	1	PIII-500	2866	2466
>>>>>
>>>>>Best Regards,
>>>>>Chris Carson
>>>>
>>>>Best regards, Wybe
>>>
>>>Rebel page:  http://www.rebel.nl/gm.htm
>>>
>>>Shows the hardware I posted.  :)
>>>
>>>Are you saying the Rebel page is wrong?
>>>
>>>Best Regards,
>>>Chris Carson
>>
>>Hi Chris,
>>  I have been keeping similar records, in Excel spreadsheet, for some time.  I
>>did not, however, record the hardware types (as yet).
>>  1. After we get the list settled, agreed by all (Rebel Team and critics), I
>>would like to capture also the color played by Rebel in each game.
>>  2. Then, I will verify the the web-based rating calculations you have posted
>>by using my own manual check.
>>  3. I already have spreadsheet calculations plus an Excel graph that show
>>Opponent Rtg, Rebel TPR (each game), Rebel running average TPR (all games),
>>Rebel moving ave TPR (last 5 games), Rebel moving ave TPR (last 10 games).
>>  4. I'd like to also graph Rebel TPR info for: A) types of Hardware, as well as
>>B) Color (White/Black results), since both are interesting stats.
>>  The calculations on types of hardware may (if enough games are played) allow a
>>test of the generally accepted principal that programs play 50-70 points better
>>for each doubling of CPU speed, to see if that principal applies to comp-human
>>play based on Rebel 40/2 tests, and what the empirical figure is in this
>>instance.
>>  Thanks for the work published so far!
>>  --Steve Boak
>
>Using all 21 games now played, I get a TPR of 2491 for Rebel. Is this what
>others are getting too? I notice above that the Grabliauskas game on the
>PIII-500 is accidently called a win for Rebel. I think it was actually a loss :(
>

It is listed as a win on the Rebel board:
http://www.rebel.nl/gm.htm

2 Virginijus Grabliauskas IM 2466 Pentium-III, 500 Mhz (white) 0 - 1

This puts the Rebel average a little higher.

If this is incorrect, get Ed to let me know and change his
information.  Whatever Ed says, I will ofcourse concur with
on this.  :)

Otherwise, your caclculations look correct.  I did include the
Hoffman as a win in my calculation above and just made a note
about the hardware.  IMHO: it was worth noting, but I will
remove the note in future posts.  :)

Remarkable score though,  especially when broken out by hardware.  :)

Best Regards,
Chris Carson

>I asked Ed at the Rebel Board whether he thought the Hoffman game should be
>included and he feels that it should be included -- Hoffman won well played game
>and Rebel's hardware problems did not cause the loss. Earlier, Ed stated here at
>CCC that he felt the Russek games should also be included since they were played
>with same settings as in GM Challenge and were played with 40/2 time control.



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.