Introducing a new branch of the Gambit Rating List

What is the strongest engine if we let them compete on 20 cores instead of one core? Can the gambit positions be better understood if we give engines 20 times more time? For this purpose we let the top-5 engines play on 20 cores and a hash table size of 4Gb. If a new engine enters the top-5 of the Gambit Rating List it will have to conquer its place in the top-5 by playing all top-5 engines and if it manages the number 5 will be removed from the list.

First run currently in progress - http://rebel13.nl/pgn4web-3.05/live-test.html

Code:

 M A T C H - I N F O

. Gambit Rating List 20 cores

. Hardware   : 2 x Intel Xeon E5 2680 v2
. Frequency  : 2.8 - 3.6 Ghz
. Used cores : 20

. Software   : Arena 3.0
. Hash table : 4Gb
. TC         : 40/120

. Openings   : gambits.pgn
. Games      : 1000
. Duration   : 2 days

. Hints
. square B8  : shows the FEN
. square C8  : shows the PGN

. Click refresh after draws to update the page.

Arena bug.

All engines are configured at 4Gb for hash tables. Following the task manager it does. But after some rounds it suddenly start to change. Now the situation is as follows:

Code:

Engine         Memory use
Rubichess      511Mb
Slowchess      485Mb
Ethereal       102Mb ????????
Komodo         527Mb
Stockfish      480Mb

I will restart the tournament tomorrow with good old cute-chess.

Restart tournament.

Code:

M A T C H - I N F O

. Gambit Rating List 20 cores

. Hardware   : 2 x Intel Xeon E5 2680 v2
. CPU        : 20 cores / 40 threads
. Frequency  : 2.8 - 3.6 Ghz
. Used cores : 20

. Software   : Cute-Chess
. Hash table : 4Gb
. TC         : 40/120

. Openings   : gambits.pgn
. Games      : 1000
. Duration   : 2-3 days

http://rebel13.nl/b/grl.htm