CEGT - Update June 13, 2010

[Wolfgang]

Hi all,

our actual rating lists are online and can be found under the attached links.

40 /120:
Actual results in our forum as usual.

40 / 20:
New games: 547
Total: 413.972

Our main 40/20-tester is on holiday, therefore we only have few new games! But this will soon be better!

NEW Engines

No new enty this week!


UPDATES

Deep Rybka 4 x64 4CPU: 3223 - 843 (+/- 0)
Deep Rybka 4 w32 1CPU: 3117 - 380 (-30) as expected , now behind x64 but only a few points. The impression that 32bit version is much more improved than the others seems to be confirmed, also by the results with 40/4 (see below).
Critter 0.70 x64 1CPU: 2979 - 410 (+7)

40 / 4:

New games: 7.290
Total: 652.484

NEW

Critter 0.70:
- x64 4CPU: 3117 - 50 Partien (first rating without comment)
- x64 1CPU: 3018 - 1398 (+124 to 0.60!!)
- w32 1CPU: 2974 - 1000 (+93 to 0.60!)

Naum 4.2 x64 2CPU: 3113 – 700

Rybka 4.0
- w32 1CPU: 3127 - 1000 (+87 to Rybka 3, -54 to x64
- x64 2CPU: 3231 - 600 (+33 to Rybka 3)

Updates
Critter 0.60 x64 2CPU: 2950 - 800 (-2)
Hiarcs 13.1 1CPU: 2951 - 2300 (-1)
Hiarcs 13.1 2CPU: 3003 - 1592 (-5)
Rybka 4.0 x64 1CPU: 3181 - 2670 (-7)
Rybka 4.0 x64 4CPU: 3264 - 1750 (-1)

A big „Thank you“ to all testers as usual!!


Links

40/20: http://www.husvankempen.de/nunn/rating.htm
Blitz: http://www.husvankempen.de/nunn/blitz.htm
40/120: http://www.husvankempen.de/nunn/rating120.htm
Tester: http://www.husvankempen.de/nunn/testers/testers.htm
Elo-comparison: http://www.husvankempen.de/nunn/Replay/ ... arison.htm

Wolfgang Battig
CEGT-Team

[Albert Silver]

Thanks for the great work.

I was intrigued by the comment on Rybka 4 32-bit, since AFAIK, the speedup of the 64-bit version over the 32-bit version is pretty much identical to that of its Rybka 3 sibling.

I am not casting doubts on the results, since we all know variance plays its role. For example, Stockfish 1.7.1 gains 48 Elo when going from 1-core to 2 cores on the list, and then 63 Elo from 2 cores to 4 cores on the same list.

[Wolfgang]

Thanks for the great work.

I was intrigued by the comment on Rybka 4 32-bit, since AFAIK, the speedup of the 64-bit version over the 32-bit version is pretty much identical to that of its Rybka 3 sibling.

I am not casting doubts on the results, since we all know variance plays its role. For example, Stockfish 1.7.1 gains 48 Elo when going from 1-core to 2 cores on the list, and then 63 Elo from 2 cores to 4 cores on the same list.

Hi Albert,

thanks for your comments and that you like our work.

I also do not understand whats going on with Rybka 4 w32 1CPU. Its only the case with 40/20, at 40/4 we have +54 from x64 to w32 which is quite "normal" (with Rybka 3 it was +80 in f/o x64).

BUT we need many more games for both with 40/20. This will be one of my tasks during next week(s)...