Re: CEGT 40/120 and 40/20 update (Juli 01st 2007)

[Werner]

Hi all ,

Our newest rating lists and games for download are now online. All links at the end of the posting as usual.

40/20
This week was focused on games with Rybka 32bit 1CPU and the newer Spiderchess version.

Rybka232 w32 1CPU
The 32bit-engine of Rybka has now 364 more games and +9 elos as last week. It´s now the best single engine - even better than Rybka 1.1 x64.

Rybka232 x64 2CPU
With 2 more points as last week Rybka 2.3.2a 2CPU is around 50 elos better than Zap!Chess Zanzibar running on 4CPUs.

Loop M1-P/M1-T 4CPU
After around 200 games the version M1-P is 38 elos better than M1-T but more games are needed. With 2CPUs the rating of these engines are equal.

SpiderChess070603
The newer version of SpiderChess shows an improvement after 179 games. We will see what happens with more games.

We made of course a lot of more games with other engines this week especially with such under 300 games in the list to get a more stable rating. An example is Chiron v.0.8.1b10 which lost 22 points and we can now see that the newer 0.87 is clear better.

40/120
This week we have a little "anniversary" because we now have made more than 10.000 games for this list! So we made an anniversary update (greetings to all Rybka and TR fans ) of our 40/120 list with the old leader. See details also here:

http://husvankempen.de/nunn/phpBB2/viewtopic.php?t=784

A big thanks to all testers as usual.

Links:
All-Versions: http://www.husvankempen.de/nunn/40_40%2 ... liste.html
Best-Versions: http://www.husvankempen.de/nunn/40_40%2 ... liste.html
Blitz: http://www.husvankempen.de/nunn/blitz.htm (also with Best-Versions and All-Versions lists)
40/120: http://www.husvankempen.de/nunn/40_120_ ... liste.html
Replay-Zone: http://www.husvankempen.de/nunn/Replay/replay.htm
Elo comparison 40/120 and 40/20: http://www.husvankempen.de/nunn/Replay/ ... arison.htm
Testers with picture and Hardware: http://www.husvankempen.de/nunn/testers/testers.html

Werner Schüle
CEGT-Team

[Oscar L]

Thanks for all your testing


Just a question. Do you have an explanation for Deep Fritz 4CPU results, I mean worse than 2CPU?

Deep Fritz 10 2CPU 2892 12 12 2053 55.2 % 2856 36.3 %

Deep Fritz 10 4CPU 2852 32 32 319 35.6 % 2955 32.9 %

Just the number of games? How many kN/s do you get with your 4CPU on the startup position?



Is is difficult to compare with CCRL, they have nearly only tested 4CPU

http://www.computerchess.org.uk/ccrl/40 ... _length=30

http://computerchess.org.uk/ccrl/404.li ... _length=30


Regards

[Werner]

Thanks for all your testing


Just a question. Do you have an explanation for Deep Fritz 4CPU results, I mean worse than 2CPU?

Deep Fritz 10 2CPU 2892 12 12 2053 55.2 % 2856 36.3 %

Deep Fritz 10 4CPU 2852 32 32 319 35.6 % 2955 32.9 %

Just the number of games? How many kN/s do you get with your 4CPU on the startup position?

Regards

Hi Oscar,
one point is statistical error - the other point is: I think there is no great difference in using 2 or 4 CPUs with Deep Fritz 10.
Of course you see more nps (8616 - 4593 in startposition after 1 minute)
But have a look for the results here: both engines reach T18 in the same time.

Werner

Neue Partie
rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1

Analysis by Deep Fritz 10 4CPU:

1.e2-e4 Sb8-c6 2.Sg1-f3 Sg8-f6 3.Sb1-c3 d7-d5 4.e4xd5 Sf6xd5 5.Lf1-b5 Sd5-f4
= (0.11) Tiefe: 9/19 00:00:00 138kN
1.d2-d4
= (0.11) Tiefe: 9/25 00:00:00 158kN
1.d2-d4 Sg8-f6 2.Sg1-f3 e7-e6 3.Sb1-d2 d7-d5 4.Sf3-e5 Sb8-d7 5.Sd2-f3 Sd7xe5
= (0.16) Tiefe: 9/25 00:00:00 196kN
1.d2-d4 Sg8-f6 2.Sg1-f3 e7-e6 3.Sb1-d2 d7-d5 4.e2-e3 Sb8-d7 5.Lf1-b5 a7-a6 6.Lb5-d3
= (0.09) Tiefe: 10/21 00:00:00 318kN
1.e2-e4
= (0.09) Tiefe: 10/22 00:00:00 365kN
1.e2-e4 Sb8-c6 2.d2-d4 Sg8-f6 3.e4-e5 Sf6-d5 4.c2-c4 Sd5-b4 5.Sb1-c3 d7-d6
= (0.14) Tiefe: 10/22 00:00:00 437kN
1.e2-e4 Sb8-c6 2.d2-d4 Sg8-f6 3.e4-e5 Sf6-d5 4.Sg1-f3 e7-e6 5.Sb1-d2 Lf8-b4 6.c2-c3
² (0.29) Tiefe: 11/31 00:00:00 1010kN
1.e2-e4 Sb8-c6 2.Sg1-f3 e7-e5 3.Lf1-c4 Lf8-c5 4.d2-d4 e5xd4 5.0-0 Sg8-f6 6.c2-c3 d4xc3
= (0.24) Tiefe: 12/30 00:00:00 1952kN
1.e2-e4 Sb8-c6 2.Sg1-f3 e7-e5 3.Lf1-c4 Lf8-c5 4.d2-d4 e5xd4 5.0-0 Sg8-f6 6.e4-e5 Sf6-e4 7.Sb1-d2 d7-d5 8.e5xd6
² (0.30) Tiefe: 13/33 00:00:00 4774kN
1.e2-e4 Sg8-f6 2.e4-e5 Sf6-d5 3.Sg1-f3 e7-e6 4.c2-c4 Sd5-b6 5.d2-d4 Lf8-e7 6.Lf1-d3 d7-d5 7.c4-c5 Sb6-c4
² (0.41) Tiefe: 14/39 00:00:02 15381kN
1.e2-e4 Sb8-c6 2.Sg1-f3 e7-e5 3.d2-d4 e5xd4 4.Sf3xd4 Sg8-f6 5.Sd4xc6 b7xc6 6.Lf1-d3 d7-d5 7.e4-e5 Dd8-e7 8.0-0 De7xe5 9.Tf1-e1 Sf6-e4
² (0.41) Tiefe: 15/35 00:00:04 37741kN
1.e2-e4 Sb8-c6 2.Sg1-f3 e7-e5 3.d2-d4 e5xd4 4.Sf3xd4 Sg8-f6 5.Sd4xc6 b7xc6 6.Lf1-d3 d7-d5 7.e4-e5 Dd8-e7 8.0-0 De7xe5 9.Tf1-e1
² (0.30) Tiefe: 16/38 00:00:07 60311kN
1.e2-e4 e7-e5 2.Sg1-f3 Sg8-f6 3.Sb1-c3 Sb8-c6 4.d2-d4 e5xd4 5.Sf3xd4 Lf8-b4 6.Sd4xc6 b7xc6 7.Lf1-d3 0-0 8.0-0 d7-d5 9.e4-e5 Sf6-g4 10.Lc1-f4
= (0.16) Tiefe: 17/40 00:00:18 155mN
1.d2-d4
= (0.16) Tiefe: 17/40 00:00:21 185mN
1.d2-d4 Sg8-f6 2.Sg1-f3 e7-e6 3.Sb1-d2 d7-d5 4.e2-e3 Sb8-d7 5.Lf1-d3 Lf8-d6 6.0-0 0-0 7.e3-e4 Sf6xe4
= (0.22) Tiefe: 17/40 00:00:26 226mN
1.d2-d4 Sg8-f6 2.Sg1-f3 d7-d5 3.e2-e3 Sb8-c6 4.Sb1-c3 Lc8-g4 5.h2-h3 Lg4-h5 6.Lf1-e2 e7-e6 7.0-0 Sf6-e4 8.Dd1-d3 Lh5-g6 9.Dd3-b5
= (0.11) Tiefe: 18/41 00:00:54 473mN
1.e2-e4
= (0.11) Tiefe: 18/41 00:00:59 512mN

(Werner, 01.07.2007)

Neue Partie
rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1

Analysis by Deep Fritz 10 2CPU:

1.e2-e4 Sb8-c6 2.Sg1-f3 Sg8-f6 3.Sb1-c3 d7-d5 4.e4xd5 Sf6xd5 5.Lf1-b5 Sd5-f4
= (0.11) Tiefe: 9/22 00:00:00 121kN
1.Sg1-f3
= (0.11) Tiefe: 9/22 00:00:00 137kN
1.Sg1-f3 Sg8-f6 2.d2-d4 e7-e6 3.Sb1-d2 d7-d5 4.Sf3-e5 Sb8-d7 5.Sd2-f3 Sd7xe5
= (0.16) Tiefe: 9/22 00:00:00 156kN
1.Sg1-f3 Sg8-f6 2.e2-e3 e7-e6 3.Lf1-e2 d7-d6 4.0-0 Sb8-d7 5.d2-d3 Lf8-e7
= (0.14) Tiefe: 10/23 00:00:00 280kN
1.e2-e4
= (0.14) Tiefe: 10/23 00:00:00 366kN
1.e2-e4 Sb8-c6 2.d2-d4 d7-d5 3.e4-e5 Lc8-f5 4.c2-c3
= (0.24) Tiefe: 10/25 00:00:00 463kN
1.e2-e4 e7-e5 2.Sg1-f3 Sg8-f6 3.Sb1-c3 Sb8-c6 4.Lf1-c4 Lf8-c5 5.0-0 0-0 6.d2-d3
= (0.22) Tiefe: 11/29 00:00:00 969kN
1.e2-e4 e7-e5 2.Sg1-f3 Sg8-f6 3.Sb1-c3 Sb8-c6 4.Lf1-b5 Lf8-c5 5.0-0 0-0 6.Lb5xc6 d7xc6 7.Sf3xe5
= (0.11) Tiefe: 12/33 00:00:00 1787kN
1.d2-d4
= (0.12) Tiefe: 12/33 00:00:00 2226kN
1.d2-d4 Sg8-f6 2.Sg1-f3 e7-e6 3.e2-e3 Lf8-e7 4.Lf1-d3 d7-d6 5.0-0 0-0 6.e3-e4 Sb8-c6
= (0.16) Tiefe: 12/33 00:00:00 2596kN
1.d2-d4 Sg8-f6 2.Sg1-f3 e7-e6 3.e2-e3 d7-d5 4.Lf1-d3 Lf8-d6 5.0-0 0-0 6.Sb1-d2 Sb8-d7 7.Tf1-e1
= (0.09) Tiefe: 13/32 00:00:01 4647kN
1.e2-e4
= (0.09) Tiefe: 13/32 00:00:01 5151kN
1.e2-e4
= (0.25) Tiefe: 13/35 00:00:01 6918kN
1.e2-e4 e7-e5 2.Sb1-c3 Sb8-c6 3.Sg1-f3 Sg8-f6 4.Lf1-b5 Lf8-b4 5.0-0 0-0 6.Tf1-e1 d7-d6 7.d2-d4 Sf6-g4
= (0.12) Tiefe: 14/36 00:00:02 11798kN
1.e2-e4 e7-e5 2.Sg1-f3 Sg8-f6 3.d2-d4 Sf6xe4 4.Lf1-d3 d7-d5 5.Sf3xe5 Lf8-b4+ 6.c2-c3 Lb4-d6 7.0-0 Sb8-d7 8.Se5-c4 Ld6xh2+ 9.Kg1xh2 Dd8-h4+
² (0.34) Tiefe: 15/38 00:00:07 33646kN
1.e2-e4 e7-e5 2.Sg1-f3 Sg8-f6 3.d2-d4 Sf6xe4 4.Lf1-d3 d7-d5 5.Sf3xe5 Lf8-b4+ 6.c2-c3 Lb4-d6 7.0-0 Sb8-d7 8.Se5-c4 Ld6xh2+ 9.Kg1xh2 Dd8-h4+
² (0.34) Tiefe: 16/39 00:00:12 56918kN
1.e2-e4 e7-e5 2.Sg1-f3 Sb8-c6 3.Lf1-b5 a7-a6 4.Lb5-c4 Sg8-f6 5.0-0 Lf8-c5 6.c2-c3 Sf6xe4 7.d2-d4 e5xd4 8.Lc4-d5 Se4-f6 9.Tf1-e1+ Ke8-f8
² (0.27) Tiefe: 17/41 00:00:26 122mN
1.e2-e4 e7-e5 2.Sg1-f3 Sb8-c6 3.Lf1-b5 a7-a6 4.Lb5-c4 Sg8-f6 5.0-0 Lf8-c5 6.c2-c3 Sf6xe4 7.d2-d4 e5xd4 8.Lc4-d5 Se4-f6
² (0.27) Tiefe: 18/42 00:00:55 255mN

(Werner, 01.07.2007)

[Oscar L]

Then may we conclude that multicore implementation is not good enough?

[Tony Thomas]

Then may we conclude that multicore implementation is not good enough?

Same goes for Shredder.