GB Championship 8CPU (completed)

[Graham Banks]

GB CHAMPIONSHIP 8CPU

Xeon X5430x2 Octal
ChessGUI
2048mb hash per engine where possible
3-4-5 piece tablebases
Ponder off
Graham2013-2.cgb book (limited to 8 move depth)
40 moves in 29 minutes repeating (equivalent to CCRL 40/40)
All engines 64-bit 8CPU where available
2 cycles 34 rounds


Participants

Houdini 3 64-bit 8CPU
Komodo 5.1 64-bit 8CPU
Stockfish 120713 64-bit 8CPU
Critter 1.6a 64-bit 8CPU
Rybka 4.1 64-bit 8CPU
Equinox 1.65 64-bit 8CPU
Bouquet 1.7b 64-bit 8CPU
Vitruvius 1.19g 64-bit 8CPU
Gull 2.1 64-bit 8CPU
IvanHoe 9.46h 64-bit 8CPU
RobboLito 0.21Q 64-bit 8CPU
Sting SF 3 64-bit 8CPU
Naum 4.2 64-bit 8CPU
Hiarcs 14 8CPU
Chiron 1.5 64-bit 4CPU
Protector 1.5.0 64-bit 8CPU
Hannibal 1.3 64-bit 8CPU
Deep Junior 13 64-bit 8CPU (con15)

[Graham Banks]

GB CHAMPIONSHIP 8CPU

Xeon X5430x2 Octal
ChessGUI
2048mb hash per engine where possible
3-4-5 piece tablebases
Ponder off
Graham2013-2.cgb book (limited to 8 move depth)
40 moves in 29 minutes repeating (equivalent to CCRL 40/40)
All engines 64-bit 8CPU where available
2 cycles 34 rounds


Final Standings

23.5 - Komodo 5.1 64-bit 8CPU
22.5 - Houdini 3 64-bit 8CPU
22.0 - Stockfish 120713 64-bit 8CPU
21.5 - Sting SF 3 64-bit 8CPU
20.0 - Critter 1.6a 64-bit 8CPU
19.5 - Equinox 1.90 64-bit 8CPU
19.0 - Rybka 4.1 64-bit 8CPU
18.5 - Bouquet 1.7b 64-bit 8CPU
17.0 - Vitruvius 1.19g 64-bit 8CPU
16.5 - IvanHoe 9.46h 64-bit 8CPU
15.5 - RobboLito 0.21Q 64-bit 8CPU
15.5 - Protector 1.5.0 64-bit 8CPU
15.0 - Gull 2.1 64-bit 8CPU
13.0 - Naum 4.2 64-bit 8CPU
12.5 - Hiarcs 14 8CPU
12.5 - Chiron 1.5 64-bit 8CPU
11.5 - Hannibal 022013 64-bit 8CPU
10.5 - Deep Junior 13 64-bit 8CPU (con15)

[lostman]

In most of the decisive games as soon as the two engines go out of the book part (eval 0.1 for both until the move
both engine almost agree that the game is lost/won immediately :
if we take the difference of evaluation of white at move 9 minus evaluation of black at move nine it is often above 50 somletimes around 100. Does it not mean that the book should be corrected and that it was because the engines followed it that they lost?
Is it possible to update the book accordingly and rerun a new turnament with the new book ?

I am still not familiar with the book making and the book usage of engines so may be it is a stupid remark sorry if it is the case.

Mehdi

[Graham Banks]

In most of the decisive games as soon as the two engines go out of the book part (eval 0.1 for both until the move
both engine almost agree that the game is lost/won immediately :
if we take the difference of evaluation of white at move 9 minus evaluation of black at move nine it is often above 50 somletimes around 100. Does it not mean that the book should be corrected and that it was because the engines followed it that they lost?
Is it possible to update the book accordingly and rerun a new turnament with the new book ?

I am still not familiar with the book making and the book usage of engines so may be it is a stupid remark sorry if it is the case.

Mehdi

Opening lines seen as unfair were auto-replayed.
The criteria used was if both sides saw the evaluation favouring one side by more than 0.69 for 14 consecutive moves once they were out of the 8-move opening book.

[lostman]

Thx for the fast reply and for the great testing and interesting pgn producing job you are doing.