This is actually a 100 game match. And I had intended to have only 1 crosstable for the 100 games.
However in the Fritz GUI, tho in the database it lists the games as 1 thru 100, in the PGN file created it lists them as 1 thru 50, and then 1 thru 50 again. I do not like that, which seems to just be an issue with engine-engine matches. Instead of changing the round numbers in the PGN file- it was just easier to list it as 2 50-game matches and edit nothing.
Between Fritz 13, Komodo 4 32bit, Rybka 4.1 32bit and Stockfish 2.2.2 32bit, it seems to be Stockfish that causes the most problems for the new Vitruvius engine. So here we go:
1CPU/32bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games
Code: Select all
1 Stockfish 2.2.2 JA +14 +14/-12/=24 52.00% 26.0/50
2 Vitruvius_1.01C_x32 -14 +12/-14/=24 48.00% 24.0/50
1 Vitruvius_1.01C_x32 +70 +19/-9/=22 60.00% 30.0/50
2 Stockfish 2.2.2 JA -70 +9/-19/=22 40.00% 20.0/50In 32bit and 1CPU, I would think that very likely this Vitruvius version is firmly entrenched in the "Top 5". And the 1 hour and longer 64bit MP games that Gabriel (Gab) has been running seem to show that it is also very likely to be in the "Top 5" there also- even giving Houdini fits.
Naturally it will take more games, especially at 64bit- and hopefully I will be running those myself by this time next week.
Hope you enjoy,
george