Test for CCRL 40/4: New engine Zorgus 1.1 64-bit

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

Sergio Martinez
Posts: 1975
Joined: Fri Oct 05, 2012 2:35 pm
Location: Spain

Test for CCRL 40/4: New engine Zorgus 1.1 64-bit

Post by Sergio Martinez »

Code: Select all

Zorgus 1.1 64-bit             :  349 (+ 91,= 76,-182), 37.0 %

Horizon 4.4                   :  30 (+  8,=  4,- 18), 33.3 %
Jumbo 0.5.3 64-bit            :  30 (+  2,=  7,- 21), 18.3 %
Eveann 1.72                   :  30 (+  3,=  6,- 21), 20.0 %
GopherCheck 0.2.3 64-bit      :  30 (+  3,= 10,- 17), 26.7 %
Isa 2.0.48 64-bit             :  20 (+  4,=  2,- 14), 25.0 %
Dorpsgek Dillinger 64-bit     :  30 (+  7,=  6,- 17), 33.3 %
Baislicka 1.0 64-bit          :  30 (+ 11,=  5,- 14), 45.0 %
Tiny Chess 1.46 64-bit        :  30 (+  7,=  8,- 15), 36.7 %
Soldat III 0.187 64-bit       :  30 (+ 11,=  6,- 13), 46.7 %
GK 0.90 64-bit                :  30 (+ 10,= 11,-  9), 51.7 %
AdroitChess 0.4               :  29 (+  8,=  7,- 14), 39.7 %
Alcibiades 0.2.7 64-bit       :  30 (+ 17,=  4,-  9), 63.3 %
New engine by Olivier de Lannoy
http://olivier.delannoy.free.fr/chess-zorgus/

PGN file: http://kirill-kryukov.com/chess/discuss ... p?id=40453
Member of the CCRL Group. Write me if you want I test your engine.
User avatar
Ras
Posts: 2773
Joined: Tue Aug 30, 2016 8:19 pm
Full name: Rasmus Althoff

Re: Test for CCRL 40/4: New engine Zorgus 1.1 64-bit

Post by Ras »

This isn't a new engine, it is NG-Play without many changes.

The UCI interface would be the only new thing and not trivial to add because NG-Play is single-threaded and sequential. Guess what, Zorgus' UCI mode doesn't accept the "stop" command during calculation.

George Georgopoulos (author of NG-Play) hasn't given a formal licence, but the header in NG-Play clearly states that while derivatives are allowed, it has to stay open source. From NG-Play source:

Code: Select all

/*         You are free to copy/derive from this code     */
/*         as long as you mention this copyright          */
/*         and keep the new code open source              */
Zorgus is violating this.

I took a few quick checks in the interactive mode with text board (looking identical!).

Position is after 1. a2-a3 a7-a6 from the initial position; this gets the engines out of the book. The following depths have exactly the same eval and principal variation up to depth 9 plies, then it becomes a bit different. (note: the CT800 engine has completely different evals and PVs).

Even the accounted move overhead is the same, 100 ms. I've checked with a hex editor, the visible function names are the same, though reading FEN positions from a file is new in Zorgus.

The biggest joke from Oliviers website:

"You are allowed to distribute Zorgus as freeware as long as the original files remain unchanged. Any commercial use is not allowed without written permission of the author."

Taking an existing open source engine, violating the "must stay open source" condition and slapping a copyright on "Zorgus"?!
Sergio Martinez
Posts: 1975
Joined: Fri Oct 05, 2012 2:35 pm
Location: Spain

Re: Test for CCRL 40/4: New engine Zorgus 1.1 64-bit

Post by Sergio Martinez »

Ras wrote:This isn't a new engine, it is NG-Play without many changes.

The UCI interface would be the only new thing and not trivial to add because NG-Play is single-threaded and sequential. Guess what, Zorgus' UCI mode doesn't accept the "stop" command during calculation.

George Georgopoulos (author of NG-Play) hasn't given a formal licence, but the header in NG-Play clearly states that while derivatives are allowed, it has to stay open source. From NG-Play source:

Code: Select all

/*         You are free to copy/derive from this code     */
/*         as long as you mention this copyright          */
/*         and keep the new code open source              */


Zorgus is violating this.

I took a few quick checks in the interactive mode with text board (looking identical!).

Position is after 1. a2-a3 a7-a6 from the initial position; this gets the engines out of the book. The following depths have exactly the same eval and principal variation up to depth 9 plies, then it becomes a bit different. (note: the CT800 engine has completely different evals and PVs).

Even the accounted move overhead is the same, 100 ms. I've checked with a hex editor, the visible function names are the same, though reading FEN positions from a file is new in Zorgus.

The biggest joke from Oliviers website:

"You are allowed to distribute Zorgus as freeware as long as the original files remain unchanged. Any commercial use is not allowed without written permission of the author."

Taking an existing open source engine, violating the "must stay open source" condition and slapping a copyright on "Zorgus"?!
Thanks for your explanation Rasmus
Member of the CCRL Group. Write me if you want I test your engine.