Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rybka EXE will bigger and bigger - why this? Imbreeding Factor

Author: Rolf Tueschen

Date: 02:48:28 01/25/06

Go up one level in this thread


On January 24, 2006 at 23:54:43, Mike Byrne wrote:

>On January 24, 2006 at 22:12:23, Rolf Tueschen wrote:
>
>>On January 24, 2006 at 22:09:27, Dann Corbit wrote:
>>
>>>On January 24, 2006 at 21:45:06, Graham Banks wrote:
>>>
>>>>On January 24, 2006 at 21:39:31, Dann Corbit wrote:
>>>>
>>>>>On January 24, 2006 at 21:34:48, Rolf Tueschen wrote:
>>>>>[snip]
>>>>>>The Old Siberian CHess wasnt bad either. But honestly isnt that a contradictio
>>>>>>in prosecco if you say strong _and_ ultra-tiny? Sounds as if you meant a
>>>>>>peabrain? How could it be strong?
>>>>>
>>>>>"I myself am often surprised at life's little quirks..." -- The man in black.
>>>>>
>>>>>Elo list from:
>>>>>http://wbec-ridderkerk.nl/
>>>>>
>>>>>RankProgram:                      Ed 11    Comm Priv Ed 10   Ed 9    Ed 8
>>>>>
>>>>>1   Fruit WCCC'05                     2794*         2631    2610    2551
>>>>>2   Shredder 9-x64                    2766*         2818    2760    2766
>>>>>3   Gandalf 6.01                      2736*         2702    2738    2582
>>>>>4   Pro Deo 1.1                       2730          2689    2672    2724
>>>>>5   Thinker 5.0b x64                  2729          2590    2575    2617
>>>>>6   Zappa 1.1 x64                     2724          2668    New     New
>>>>>[snip]
>>>>>
>>>>>The 5.0b version is not generally available (I understand he is working on an
>>>>>update now) but the version that I have is also very strong and is 70K in size.
>>>>
>>>>
>>>>The Thinker.exe file for Thinker 5.0b is 49kb in size!
>>>
>>>If we continue our extrapolation, it becomes infinitely smart at zero size.
>>>;-)
>>
>>Also it couldnt be engineered backwards...
>>
>>But you guys please tell me why isnt THinker promoted and supported like Rybka?
>>I dont get it.
>
>Simple - There is nothing else like Rybka -- it is  - at the moment, the
>strongest engine per Ghz.
>
>The endefeated Rybka is the standard Beta 12, all others have tweaked settings.
>
>Many more rounds to play.
>
>    Engine                            Score
>01: Rybka v1.01 Beta 12.x64           5.0/5
>02: Rybka v1.01 Beta 12.x64 [001]     4.0/5
>03: Rybka v1.01 Beta 12.x64 [111]     4.0/5
>04: Rybka v1.01 Beta 12.x64 [010]     4.0/5
>05: Pro Deo                           4.0/5
>06: Toga II                           3.5/5
>07: Rybka v1.01 Beta 12.x64 [011]     3.5/5
>08: Rybka v1.01 Beta 12.x64 [110]     3.5/5
>09: GFruit-4bx                        3.0/5
>10: Shredder 8                        3.0/5
>11: Fruit 2.1                         3.0/5
>11: Rybka v1.01 Beta 12.x64 msvtus6vp 3.0/5
>13: Shredder Classic 1.3              3.0/5
>14: Rybka v1.01 Beta 12.x64 [101]     3.0/5
>15: Rybka v1.01 Beta 12.x64 [100]     3.0/5
>16: Rybka v1.01 Beta 12.x64 MFB       2.5/5
>17: Rybka v1.01 Beta 10d.x64 MFB      2.5/5
>18: Shredder 9.1                      2.0/5
>18: Rybka v1.0 Preview 2 x64          2.0/5
>20: Gandalf60                         2.0/5
>21: Crafty64 19.20                    2.0/5
>22: Rebel 12                          1.5/5
>23: Deep Sjeng 1.6                    1.0/5
>24: AnMon 5.50                        0.5/5
>25: Zappa 1.1                         0.5/5
>25: Ruffian 2.1                       0.5/5
>27: TheKing 3.23                      0.5/5
>28: SOS 5.1 for Arena                 0.0/5



Mike,

what you oversee is this: you could get the same results if you only had a
single Rybka entry and some 20 actually differently tuned Shredders!! This is
stats and NOT original strength. I used to give the example of the Bobby Fischer
problem in 1962 Curacao... You have no chance as a single player if some 6 or 8
opponents can enter cooperation mode just by their mere number of players.



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.