Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: SSDF, Programs, and GM's

Author: Chris Carson

Date: 07:42:10 02/08/00

Go up one level in this thread


On February 08, 2000 at 09:41:30, Albert Silver wrote:

>On February 08, 2000 at 08:40:44, Chris Carson wrote:
>
>>Here is some interesting data that supports that the SSDF is a
>>valid measure of both Comp vs Comp and Comp vs Human strength (FIDE)
>>and that the programs are playing at GM strength:
>>
>>                                        Opp
>>Time Control 40/2                       FIDE
>>Prog	Mp    	MHZ 	TPR 	SSDF	Avg	+ 	=	-	Tot
>>R9	Pmmx	225	2683	2521	2317	4	1	0	5
>>G4	Pmmx	133	2518	2468	2245	2	1	0	3
>>H6	Pmmx	200	2500	2518	2500	0	2	0	2
>>N3.5	Pmmx	133	2499	2429	2306	1	1	0	2
>>R8	Pmmx	200	2325	2513	2485	1	0	2	3
>>F4	Pmmx	133	2306	2494	2306	1	0	1	2
>>H6	Pmmx	133	2306	2493	2306	1	0	1	2
>>Mcp7	Pmmx	133	2306	2478	2306	1	1	1	3
>>R8	Pmmx	133	2306	2513	2306	0	1	0	1
>>						11	7	5	23
>>
>>                                        Opp
>>AEGON 97 (Time Controls 40/90 Fisher)   FIDE
>>Prog	Mp    	MHZ 	TPR 	SSDF	Avg	+	=	-	Tot
>>F4	Ppro	200	2743	2560	2470	1	2	0	3
>>R9	Ppro	200	2651	2553	2458	4	1	1	6
>>CM5K	Ppro	200	2606	2487	2413	2	2	0	4
>>N3.5	Ppro	200	2573	2590	2424	3	1	1	5
>>S2	Ppro	200	2458	2552	2458	1	0	1	2
>>H6	Ppro	200	2363	2559	2363	1	1	1	3
>>G4	Ppro	200	2357	2534	2429	1	2	2	5
>>						13	9	6	28
>>
>>Comments:
>>For the 40/2 results:  4 TRP's above SSDF and 5 TPR's below SSDF
>>For Aegon 97 results:  3 TPR's above SSDF and 4 TPR's below SSDF
>>No real difference between 40/2 and Aegon 97 results compared to SSDF
>>
>>Also note that excluding the Aegon 97 results (using only 40/2 games),
>>Rebel has a 20 consequetive game (needed to establish a FIDE rating)
>>rating of 2595 (this is equivelant to FIDE #100 rated player).  The
>>average GM rating is 2524 and this Rebel rating would be on 200MHZ to
>>400MHZ hardware.
>>
>>It is also worth noting that both the 40/2 and the Aegon 97 time
>>controls are valid for FIDE ratings.  Rebel has two TPR's above 2650
>>(FIDE) on this HW.
>>
>>Best Regards,
>>Chris Carson
>
>Which Rebel played 20 consecutive games at a performance of 2595? What were the
>conditions, and what was the time-control? Also, I don't think one compare the
>Hiarcs 6 rating (for example) on the SSDF, played over hundreds of games at
>40/2, with only 2 against humans.
>
>                                       Albert Silver

Here are 20 consequetive games at 40/2 that give a rating of 2553,
another very strong (above GM average of 2524):

Nov-97	3rd Copa Santos		2190	Reb 9	P-225		1
Nov-97	3rd Copa Pena		2330	Reb 9	P-225		1
Nov-97	3rd Copa Dominguez	2340	Reb 9	P-225		0.5
Nov-97	3rd Copa Mateo		2408	Reb 9	P-225		1
Nov-97	3rd Copa Hernandez	2320	Reb 9	P-225		1
Dec-98	4th Copa Dominguez	2340	Reb 10	PII-400		0.5
Dec-98	4th Copa Pinal		2330	Reb 10	PII-400		1
Dec-98	4th Copa Alvarado	2180	Reb 10	PII-400		1
Dec-98	4th Copa Fernendez	2180	Reb 10	PII-400		1
Dec-98	4th Copa Mateo		2415	Reb 10	PII-400		1
Dec-98	4th Copa Montes		2130	Reb 10	PII-400		0.5
Dec-98	4th Copa Jaquez		2220	Reb 10	PII-400		1
May-99	RGMC	 Rohde		2585	Reb-Cen	K6II-450	0
Jul-99	RGMC	 Sorin		2524	Reb-Cen	K6III-600	0.5
Jul-99	RGMC	 Anand		2795	Reb-Cen	K6II-450	0.5
Jul-99	RGMC	 Anand		2795	Reb-Cen	K6II-450	0
Aug-99	RGMC	 Rohde		2585	Reb-Cen	K6III-600	0.5
Sep-99	RGMC	 Hoffman	2501	Reb-Cen	K6III-600	0
Sep-99	RGMC	 Zapolskis	2399	Reb-Cen	K6II-300	1
Sep-99	RGMC	 Ruzele		2515	Reb-Cen	K6III-600	0.5

Note: The last 8 games were match play and one loss (Hoffman) was
due to hardware failure.  Also, I am not sure this is the highest
20 consequesitive games, but it is a second example of a very strong
(above GM average) performance by Rebel.  :)

Best Regards,
Chris Carson



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.