Computer Chess Club Archives


Search

Terms

Messages

Subject: CCRL 40/40 Rating list and stats updated

Author: Kirill Kryukov

Date: 08:49:55 02/19/06


Hi all! I am happy to announce the weekly update of CCRL rating list and
statistics:

http://ccrl.computerchess.org.uk/4040/

Please notice the new address (and update your bookmarks). Our original server
was overloaded with visitors after our previous announcements, it was getting
slow and sometimes even not responding. That's why we decided to switch to
professional hosting. Ray Banks was so generous to donate a 1 year hosting plan
to our project, so now we should have it accessible about 99.97% of time. The
new server is located in UK which should make browsing much faster for European
readers (previous one was in Japan).

This week we are adding 803 new games in CCRL 40/40 time control (benchmarked to
Athlon 64 3800+), thanks to enthusiastic efforts of our testers: Graham Banks,
Ray Banks, Sarah Bird, Kirill Kryukov, Tom Logan, Andreas Schwartmann, Charles
Smith and Chris Taylor.

So, what happened this week?

Three new engines have their first games in our database: Spike 1.1, Naum 1.91
32-bit and 64-bit. You can see their current rating in the "All versions" rating
list, but please wait with and serious discussion until they get more games.

The most interesting thing is perhaps Hiarcs 10 Hypermodern - it got 180 new
games this week, so it can now enter "Best versions" list and "Pure" list.
Unfortunately with those games it lost 18 points in our complete list, so Hiarcs
10 is now still listed in the "Best versions" list. However in the "Pure"
context, Hiarcs 10 Hypermodern could manage to keep lead over Default, so it is
now listed in our "Pure list":

http://ccrl.computerchess.org.uk/4040/rating-table-pure.shtml

(Please read the explanation at that page to see how it is possible)

A little news on correlation statistics. Following advice of Uri Blass (IIRC),
the correlation tables now have "minimum number of moves" requirement. All
engine pairs that have correlation based on less than 300 moves are shown
without special colors now. It means that only more or less reliable results now
have color in that tables. More changes will follow, including detailed
explanation about construction of those tables.

I almost forgot to mention, that today we are celebrating out two months
anniversary. It was December 19-th when first CCRL games were played, according
to our database. Two months of hard work and active discussion, two months of
friendly atmosphere and inspiration by the results. I hope that the project
lives long enough to see a 3000 ELO engine, and then 3100, 3200... That is what
I am looking forward to. :-)

As usual our forum is open for those curious about our study:

http://kd.lab.nig.ac.jp/chess/discussion-board/viewforum.php?f=7

Please feel free ask any questions or suggest new ideas. (Team-mates, please add
all that I forgot to say).

Best,
Kirill



This page took 0.02 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.