Dorpsgek Eve's Temptation 64-bit Gauntlet for CCRL 40/40

[tpoppins]

Thank you for the correction, Sven.

So this

The combined error margins will be close to 100.

should read instead "will be around 55" (for the 128 and 323 games quoted by Carlos)?

[tpoppins]

Code: Select all

CCRL 40/40 Rating List - Custom engine selection
816025 games played by 2140 programs, run by 21 testers
Ponder off, General books (up to 12 moves), 3-4-5 piece EGTB
Time control: Equivalent to 40 moves in 40 minutes on Athlon 64 X2 4600+ (2.4 GHz), 
about 15 minutes on a modern Intel CPU.
Computed on April 5, 2018 with Bayeselo based on 816'025 games
Tested by CCRL team, 2005-2018, http://computerchess.org.uk/ccrl/4040/

              Engine                  Elo   +    -   Score  AvOp  Games
  Dorpsgek Dillinger 64-bit          2202  +21  -21  49.2%   +4.7   790
  Dorpsgek Eves-Temptation 64-bit    2200  +26  -26  51.5%  -10.8   525

After two 320-game gauntlets for each version (same opponents as in the first post of this thread) we get this:

Code: Select all

CCRL 40/40 Rating List - Custom engine selection
819589 games played by 2137 programs, run by 21 testers
Ponder off, General books (up to 12 moves), 3-4-5 piece EGTB
Time control: Equivalent to 40 moves in 40 minutes on Athlon 64 X2 4600+ (2.4 GHz), 
about 15 minutes on a modern Intel CPU.
Computed on April 14, 2018 with Bayeselo based on 819'589 games
Tested by CCRL team, 2005-2018, http://computerchess.org.uk/ccrl/4040/

              Engine                  Elo   +    -   Score  AvOp  Games
  Dorpsgek Eves-Temptation 64-bit    2206  +21  -21  52.2%  -16.5   845
  Dorpsgek Dillinger 64-bit          2192  +18  -18  48.6%  +10.1  1130

with 83.7% LOS.

It's worth noting that the Dillinger version had 17 stalls and disconnects during the above-mentioned gauntlet, while the Eve's Temptation had none at all in all three tests. That alone makes the new version worthwhile.

[Sven]

Thank you for the correction, Sven.

So this

The combined error margins will be close to 100.

should read instead "will be around 55" (for the 128 and 323 games quoted by Carlos)?

I don't know, simply because I do not understand the meaning of columns in the table given by Carlos. They do not appear to come from BayesElo which is the only rating program that I am familiar with, and there are no column headings so I only partially understand their meaning. Where does your "55" result from?