New engine releases & news H1 2026

[Carbec]

Hi,

Zangdar dev version down to 171 Mo during a bench (with 128 Mo Hash).
That why we need testers , thanks to them ))

Philippe

[Frank Quisinsky]

Hi there,

updates from the last minutes (one in the list), days ...

Shallow Guess 1.0.0
https://github.com/buildingwheels/ShallowGuess
by Zixiao Han, China

Prune 3.2.1
Prune 3.2.1 HCE
Prune 3.2.0
Prune 3.2.0 HCE
https://github.com/tgirolami09/Prune
by T. GIROLAMI, J. BIENVENUE, France

Malika 0.892
https://github.com/FauziAkram/Malika-releases
by Fauzi Akram Dabat, Jordan

RuthIron 21.9.5.2
https://github.com/themreKalkan/RuthIron
by Emre Kalkan, Turkey

Rusty-Rival 1.0.29 ... and newer RC versions, current = 1.0.30 RC6
https://github.com/chris-moreton/rusty-rival/tags
by Chris Moreton, England

Anodos 1.3.0
https://github.com/tomcant/anodos
Tom Cant, England

Best
Frank

[Frank Quisinsky]

Hint:
Different engines require:
libgcc_s_seh-1.dll, libwinpthread-1.dll, libgomp-1.dll

Updated:
https://github.com/niXman/mingw-builds-binaries
Can be found under: x86_64-posix-seh-msvcrt-rt

Best
Frank

[Graham Banks]

Hi,

Zangdar dev version down to 171 Mo during a bench (with 128 Mo Hash).
That why we need testers , thanks to them ))

Philippe

Is there a fixed version yet, Philippe?

[Carbec]

Hi Graham,

Indeed, the corrected version is up.
172 Mo in play with one thread !

https://github.com/Carbecq/Zangdar/releases/tag/v6.1.1

[Graham Banks]

Hi Graham,

Indeed, the corrected version is up.
172 Mo in play with one thread !

https://github.com/Carbecq/Zangdar/releases/tag/v6.1.1

[Frank Quisinsky]

Hi Philippe,

the question is ...
Have that any effects on playing strength?

I saw that too during my test-run but I gave it not a high priority. System have 64Gb Ram and no problems during 16 matches, runs to the same time without ponder, 5-pieces and 256Mb for Hash.

More a cosmetic flaw?

Best
Frank

[Carbec]

Hi Frank

Its not a cosmetic flaw, but a real problem. I didn't encountered it until I made a test with 32 threads. Then I got a stack overflow. So people with less memory will crash. I had to change the way I allocate memory in the code.
By the way I did a quick sprt test against the 6.1 version, and won some little elo ).
By default Zangdar takes now 170 Mo for 1 thread.
I hope its ok now.

Friendly
Philippe

[Frank Quisinsky]

Hi Philippe,

OK, in this case I will start tommorrow a new test.
Thanks for your bugfix!

Best
Frank

[Frank Quisinsky]

Code: Select all

                               points   1:0 1/2  0:1     %   m-avg     1/0<50       rk
--------------------------------------------------------------------------------------
03. Zangdar 6.1                 927,0   632 590   78   71,31   81   13/  2  + 11   07#
--. Zangdar 6.1.1               925,0   618 614   68   71,15   83   17/  1  + 16   07#

1/0<50 = 1:0 / 0:1 under 50 moves
rk = ranks ... 07 = Lieutenant Colonel
Elo is boring!

I made that all in LIVE mode on my engine-overview:
https://www.amateurschach.de/main/_engines.htm

1.300 games vs. 26 opponents ...
I think around all is the same, for me the bug from 6.1 is not really important, because my test-system have 64Gb RAM.

Interesting is:
More fast won games and lesser lost games.
Move average is a bit higher ...
But all in all ... with only 1.300 games, this is all within the margin of error.

Fact is ...
Zangdar is really very strong!!

Strongest French engine I think but I have to test Gaia and I am not ready with the group of French engines!
Zangdar proudly holds the French flag.

Keep up your good work!
Next step is Colonel before General with one *



Best
Frank