Testing engines R3, N4 & Stockfish 1.3 and books.

[Tomcass]

After my first tests showing a very good performance for Stockfish 1.3 JA I have decided to test it against Rybka3 and Naum4, with three different books, Magnificent4, CompMaster Beta and Perfect15 , with 40 games each engine with each book.

The result: Rybka3 has been the best one, as expected, outperforming Naum4 for 9 points out a total of 120 games. Stockfish 1.3 JA has got an excellent -imho- 30,8% against such powerful opponents.

Among the books, CompMaster Beta has been a positive surprise, performing slightly better than my favourite until now: Magnificient4. Perfect15 very far from them. All the results here:

2009-05Stockfish(4)13-4 2009
Fritz 12 GUI
Games in 10 minutes
Quad 6700

TEST 1
Naum 4 (x4) with CompMaster Beta book
Rybka 3 ( x4) with Perfect 15 book.
Stockfish 1.3 JA with Magnificient 4 book

1 2 3
1 Naum 4(x4) 12.0 - 8.0 14.0 - 6.0 26.0/40
2 Rybka 3 32-bit (x4) 8.0 - 12.0 11.0 - 9.0 19.0/40
3 Stockfish 1.3 JA 6.0 - 14.0 9.0 - 11.0 15.0/40

TEST 2
Naum 4 (x4) with Perfect 15 book
Rybka 3 ( x4) with Magnificient 4 book
Stockfish 1.3 JA with CompMaster Beta book.

2009-05Stockfish(5)13 2009

1 2 3
1 Rybka 3 32-bit (x4) 15.0 - 5.0 13.5 - 6.5 28.5/40
2 Naum 4(x4) 5.0 - 15.0 11.5 - 8.5 16.5/40
3 Stockfish 1.3 JA 6.5 - 13.5 8.5 - 11.5 15.0/40

TEST 3

Naum 4 (x4) with Magnificient 4 book
Rybka 3 ( x4) with CompMaster Beta book
Stockfish 1.3 JA with Perfect 15 book.

2009-05Stockfish(6)13 2009

1 2 3
1 Rybka 3 32-bit (x4) 11.5 - 8.5 17.0 - 3.0 28.5/40
2 Naum 4(x4) 8.5 - 11.5 16.0 - 4.0 24.5/40
3 Stockfish 1.3 JA 3.0 - 17.0 4.0 - 16.0 7.0/40

FINAL RESULTS
1.- RYBKA 3 32-BIT (X4) 76/120 (63,33%)
2.- NAUM 4 32-BIT (X4) 67/120 (55,83%)
3.- STOCKFISH 1.3 JA (X4) 37/120 (30,83%)

FINAL RESULTS FOR BOOK TESTING.
1.- COMPMASTER BETA 69,5/120 (57,91%)
2.- MAGNIFICIENT 4 68/120 (56,66%)
3.- PERFECT15 42,5/120 (35,42%)

Regards from Barcelona,

Tom Casanovas.

[Eelco de Groot]

Thanks Tom!

That is a very interesting tournament! It is not really that the numbers can be translated to elo-ratings but it is more about the general impression you get, even just from your tables and comments; you have been watching some of the games so your observations are, I'm sure, helpful to the programmers. And also it is not yet another game in 1 second tournament

I'm not involved with Stockfish programming but its code is very interesting and am secretly tweaking the codes a bit, I hope its programmers don't mind... So I'm sidewise interested in what the results are for Stockfish, in the ratinglists, and these tournaments too to compare...

Eelco

[mcostalba]

Thanks Tom!
I'm not involved with Stockfish programming but its code is very interesting and am secretly tweaking the codes a bit, I hope its programmers don't mind...

Its programmers are very happy about this


This is the main reason why its programmers are involved in its developing

[Dr.Wael Deeb]

Thanks Tom!
I'm not involved with Stockfish programming but its code is very interesting and am secretly tweaking the codes a bit, I hope its programmers don't mind...

Its programmers are very happy about this


This is the main reason why its programmers are involved in its developing

Marco,one comment....you are a great guy
Dr.D

[Eelco de Groot]

Thanks Marco! Doc, I absolutely agree with you