Komodo 4 running for the IPON

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

IWB
Posts: 1539
Joined: Thu Mar 09, 2006 2:02 pm

Komodo 4 running for the IPON

Post by IWB »

Running for a couple of hours already:

http://www.inwoba.de

Have fun
Ingo
Bram Visser
Posts: 52
Joined: Wed Oct 19, 2011 3:37 pm
Location: NL

Re: Komodo 4 running for the IPON

Post by Bram Visser »

Funny results for now. Komodo has > 50% against all engines, but a lower rating than Houdini ...
IWB
Posts: 1539
Joined: Thu Mar 09, 2006 2:02 pm

Re: Komodo 4 running for the IPON

Post by IWB »

Bram Visser wrote:Funny results for now. Komodo has > 50% against all engines, but a lower rating than Houdini ...
Yes, looks funny, but the average of Houdini vs the opoonents is higher than the 77% of Komodo ... and of course it is not over yet!

Bye
Ingo

PS: And of course the real rating with Bayeselo will be slightly different than what is shown "on the fly".
Uri Blass
Posts: 11121
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Komodo 4 running for the IPON

Post by Uri Blass »

The average of komodo is clearly higher than the rating that it gets.

Komodo 4 SSE42 - Houdini 2.0 STD (3022) 45.5 - 44.5 50.56% Perf=3025
Komodo 4 SSE42 - Deep Rybka 4.1 SSE42 (2956) 46.0 - 44.0 51.11% Perf=2963
Komodo 4 SSE42 - Critter 1.2 (2953) 49.5 - 40.5 55.00% Perf=2987
Komodo 4 SSE42 - Stockfish 2.1.1 JA (2941) 48.5 - 40.5 54.49% Perf=2972
Komodo 4 SSE42 - Chiron 1.1a (2833) 58.5 - 30.5 65.73% Perf=2946
Komodo 4 SSE42 - Naum 4.2 (2827) 62.0 - 28.0 68.89% Perf=2965
Komodo 4 SSE42 - Deep Shredder 12 (2800) 61.0 - 28.0 68.54% Perf=2935
Komodo 4 SSE42 - Gull 1.2 (2796) 66.5 - 23.5 73.89% Perf=2976
Komodo 4 SSE42 - Deep Sjeng c't 2010 32b (2788) 71.0 - 19.0 78.89% Perf=3017
Komodo 4 SSE42 - Spike 1.4 32b (2784) 69.5 - 20.5 77.22% Perf=2996
Komodo 4 SSE42 - Protector 1.4.0 (2760) 69.0 - 20.0 77.53% Perf=2975
Komodo 4 SSE42 - Hannibal 1.1 (2757) 72.0 - 18.0 80.00% Perf=2997
Komodo 4 SSE42 - spark-1.0 SSE42 (2757) 76.5 - 13.5 85.00% Perf=3058
Komodo 4 SSE42 - HIARCS 13.2 MP 32b (2751) 78.0 - 10.0 88.64% Perf=3107
Komodo 4 SSE42 - Deep Junior 12.5 (2732) 75.0 - 15.0 83.33% Perf=3011
Komodo 4 SSE42 - Zappa Mexico II (2717) 78.5 - 10.5 88.20% Perf=3066
Komodo 4 SSE42 - Deep Onno 1-2-70 (2685) 80.0 - 8.0 90.91% Perf=3085
Komodo 4 SSE42 - Strelka 2.0 B (2673) 81.5 - 8.5 90.56% Perf=3065
Komodo 4 SSE42 - Umko 1.2 SSE42 (2664) 76.0 - 12.0 86.36% Perf=2984
Komodo 4 SSE42 - Loop 2007 (2621) 79.0 - 11.0 87.78% Perf=2963
Komodo 4 SSE42 - Jonny 4.00 32b (2614) 81.5 - 8.5 90.56% Perf=3006
Komodo 4 SSE42 - Tornado 4.80 (2609) 82.0 - 7.0 92.13% Perf=3036
Komodo 4 SSE42 - Crafty 23.3 JA (2599) 82.5 - 6.5 92.70% Perf=3040
1589.5 - 467.5 77.27% Perf=2979




2057 out of 2300 games played
The average based on this data is
(3025+2963+2987+2972+2946+2965+2935+2976+3017+2996+2975+2997+3058+3107+3011+3066+3085+3065+2984+2963+3006+3036+3040)/23=69175/23=3007.609>2979

I understand that I cannot expect total equality in the average because for extreme cases like case that the program has 100% or almost 100% against another program we are not going to get a logical result but
we have no extreme cases and the highest performance is 3107 so I do not understand what is the reason that komodo does not get at least rating of 3000.
IWB
Posts: 1539
Joined: Thu Mar 09, 2006 2:02 pm

Re: Komodo 4 running for the IPON

Post by IWB »

Hi Uri,

Yea, we had this dicsussion before with other engines. The "on the fly" Elo are calculated with the overall percentage against the average elo of the opponents and not with the average performance vs each engine.

We will see in about 3h what Bayeselo will make out of it. Bayes is my final instance for the decision. Everything else is just to have a picture. Nonetheless I think it will not be that much different from the current estimation.

Bye
Ingo
MM
Posts: 766
Joined: Sun Oct 16, 2011 11:25 am

Re: Komodo 4 running for the IPON

Post by MM »

Bram Visser wrote:Funny results for now. Komodo has > 50% against all engines, but a lower rating than Houdini ...
Yes, more than 40 elo behind....

Regards
MM
MM
Posts: 766
Joined: Sun Oct 16, 2011 11:25 am

Re: Komodo 4 running for the IPON

Post by MM »

For the 1st time, Houdini loses a match againt another engine...

http://www.inwoba.de/


Regards
MM
Uri Blass
Posts: 11121
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Komodo 4 running for the IPON

Post by Uri Blass »

MM wrote:
Bram Visser wrote:Funny results for now. Komodo has > 50% against all engines, but a lower rating than Houdini ...
Yes, more than 40 elo behind....

Regards
I do not believe it is more than 40 elo behind and I guess something is wrong with the rating that is written because komodo has many performances above 3000 based on the results.

I expect Houdini to be number 1 in the list but
I expect better result for komodo.
IWB
Posts: 1539
Joined: Thu Mar 09, 2006 2:02 pm

Re: Komodo 4 running for the IPON

Post by IWB »

Testrun finished.

Have a look at http://www.inwoba.de for results.

Bye
Ingo
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: Komodo 4 running for the IPON

Post by Laskos »

Uri Blass wrote:
MM wrote:
Bram Visser wrote:Funny results for now. Komodo has > 50% against all engines, but a lower rating than Houdini ...
Yes, more than 40 elo behind....

Regards
I do not believe it is more than 40 elo behind and I guess something is wrong with the rating that is written because komodo has many performances above 3000 based on the results.

I expect Houdini to be number 1 in the list but
I expect better result for komodo.
I am somehow surprised by the 2979 Bayeselo result. Could someone use EloStat to compare?

Kai