Rybka 4 vs Stockfish 1.8 - Long TC Match - LIVE BROADCAST!

[Martin Thoresen]

_
_
Starts Saturday July 10th at 14:00 CET
_
_


Time Control:
120 Minutes For The First 40 Moves
60 Minutes For The Next 20 Moves
15 Minutes For The Rest
+30 Seconds Added Per Move From Move 61

Misc:
Opening Database: AH-48_Basic_Positions.pgn (24 Positions Used)
CPU: Intel Core i7 920 @ 4050 MHz
OS: Windows 7 Professional x64
Tablebases: 3-4-5 Men Nalimov
Match Length: 48 Games
GUI: Deep Fritz 12
Hash: 2048 MB
Ponder: Off


Live Broadcast

Click Here - Updated Each ~3 Minutes

A full PGN will be available after the match has ended,
meanwhile you can view the games in the PGN-viewer on the
website as the match progresses.

[Martin Thoresen]

The match has started, 30 minutes earlier than planned!

Go Stockfish!

[IGarcia]

Hi Martin

Tnx for hosting. Ill be following with big interest.

Rybka stings are default ones or some custom settings? I ask because some test are done using special stings, for example in CCRL you can see "Rybka 4 C13510":

Code: Select all

Rybka 4 C13510 64-bit	#1  (3315	+27 −27	)	Quote 	
Author: Vasik Rajlich 	Link: Homepage
Rybka 4 with the following settings changes White Pawn cp=1 White Knight cp=3 White King Bishop cp=3 White Queen Bishop cp=3 White Rook cp=5 White Queen cp=10

Me thinks is it unfair because they are tunning Rybka but they not tunning Stockfish or other engines... IMHO, should be "out of the box settings" for all engines, settings that engine authors sets as best for a general default setting.

[Uri Blass]

I see nothing unfair.

Every tester is free to test the setting that he likes to test.
I see no reason to limit testers to default setting.

If you want to test different setting than the default setting for stockfish you can join CCRL and test.

Uri

[IGarcia]

I see nothing unfair.

Every tester is free to test the setting that he likes to test.
I see no reason to limit testers to default setting.

If you want to test different setting than the default setting for stockfish you can join CCRL and test.

Uri

Every body is free to see what ever wants. I see unfair.

You are right on freedom for testers. That is ok, set conditions (like time per moves, hash, tablebases, etc) equal for all. But if you start tuning engines, try to tune all of them, no one in particular. If not, "for me" its unfair. you think what ever like to think.

My language is not English native, I'm Not trying to argue. I'm just asking Marting if he is using default Rybka settings or not. In case is using custom Rybka settings I will not longer follow the match because " for me" the test is useless and unfair. You, my friend, do think and see what ever you like to do think or see

[S.Taylor]

The match has started, 30 minutes earlier than planned!

Go Stockfish!

Go both!
It's not about strength testing, it's only for entertainment!

[S.Taylor]

I think Rybka had sf on the ropes, withOUT 38.BxN. (G2)

[S.Taylor]

SF almost saved this game (G2), which would have shown greatness.
My (early) prediction is that SF will win atleast 1.5 more games from 48 than Houdini did. (but hopefully more) Unless, SF has a problem in choosing wrong endgames.

In this game, after all the terrible attack and near mates SF was faced with, the only thing wrong at the end, when it had all been thwarted, was that Rybka could get his pawns at the end of it, which means he could then march his own down to queen. I don't know if SF had another way withOUT letting Rybka do this at the end of it.

[IGarcia]

Games are very fun. G2 end was not fine for SF1.8 Hope it gets better.

I will be very happy, as an open-source fan, to see SF winning. But it will not happen now. Its just matter of time. By now, any result close to Houdini manage to get against R4 will be very good.

[gerold]

SF almost saved this game (G2), which would have shown greatness.
My (early) prediction is that SF will win atleast 1.5 more games from 48 than Houdini did. (but hopefully more) Unless, SF has a problem in choosing wrong endgames.

In this game, after all the terrible attack and near mates SF was faced with, the only thing wrong at the end, when it had all been thwarted, was that Rybka could get his pawns at the end of it, which means he could then march his own down to queen. I don't know if SF had another way withOUT letting Rybka do this at the end of it.

Thanks for you remarks Taylor.
In fast TC Houdini beats up on SF.
Houdini in fast TC is about equal with R4.

In Long TC it is yet to be decided. Thanks to Martin for running
these matches.

Best,
Gerold.

P.S. A post above mention one engine being tuned. I am hoping
both are being run with out of the box settings.