How are latest NCM Stockfish results explained?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Jouni
Posts: 3279
Joined: Wed Mar 08, 2006 8:15 pm

How are latest NCM Stockfish results explained?

Post by Jouni »

After 2018-10-23 suddenly +200 to SF7 and rating now stays there! But it was only: Revert Pull Request #1771 no functional change. Was "Correctly track down pv even in fail-high case" actually big regression? Or was it changed bench: Time controls are set to 30+0.3 and scaled periodically by comparing the server's Stockfish bench score to 1600k nps? Why is this scaling used at all?
Jouni
Top
Jouni
Posts: 3279
Joined: Wed Mar 08, 2006 8:15 pm

Re: How are latest NCM Stockfish results explained?

Post by Jouni »

There it is: " Notice: NCM upgraded all servers to to Ubuntu 18.04 (from 16.04) on October 23rd. Binaries compiled on the upgraded disto are skewing measurements upwards. We are working on a fix." Just replay 300 000 games or 3 000 000 games :P .
Jouni
Top
MikeGL
Posts: 1010
Joined: Thu Sep 01, 2011 2:49 pm

Re: How are latest NCM Stockfish results explained?

Post by MikeGL »

Jouni wrote: ↑Mon Nov 12, 2018 2:45 pm There it is: " Notice: NCM upgraded all servers to to Ubuntu 18.04 (from 16.04) on October 23rd. Binaries compiled on the upgraded disto are skewing measurements upwards. We are working on a fix." Just replay 300 000 games or 3 000 000 games :P .
Looks like Ubuntu 18.04 had a strange undocumented feature. ; )
Hope they won't blame systemd for that. :oops:
I told my wife that a husband is like a fine wine; he gets better with age. The next day, she locked me in the cellar.
Top

Return to “Computer Chess Club: General Topics”