New engine releases 2020

[AndrewGrant]

Andreas prefers NNUE as default.

Then why is 'use NNUE ' false by default?

For the event where the binary is not accompanied by a network, possible?

[Frank Quisinsky]

Hi there,

since yesterday a new Minic:

v3.01
https://github.com/tryingsomestuff/Minic
https://github.com/tryingsomestuff/Minic-Dist/
https://github.com/tryingsomestuff/NNUE-Nets

Have thanks to the chief inspector of Minic ... Vivien Clauzon!
Works fine with NeNu, like it.

Best
Frank

[Frank Quisinsky]

Hi there,

the programmer of Orion have a plan, an idea?

v0.8

Orion by David Carteau
https://orionchess.pagesperso-orange.fr/

A lot of French power to the same time!

Best
Frank

[RubiChess]

Andreas prefers NNUE as default.

Then why is 'use NNUE ' false by default?

Okay, there is some talk about latest Rubi here so I give some answers:

1. I'm not too proud about Rubi now being another NNUE player so I decided not to make an anouncement here.
2. I did a release on my github cause people already started to use some unofficial (and badly compiled) binaries spread on some file hosters for their ranking lists.
3. "Use NNUE" is set to false by default because of 1. and because I want to be able to test at OpenBench in the future and wasn't sure if I can set options in the test setup.
4. In general I will leave it to the users to use Rubi with internal evaluation or using the NNUE but Graham has set Rubi to play his Div1 and there even with adaption of the Cfish NNUE eval code (different to most other engines Rubi has a different piece representation and needs some adaption) and the usage of SF trainer to generate a network from own training data Rubi is probably still more original in Div1 than at least five contenders.

Regards, Andreas

[Pi4Chess]

Andreas prefers NNUE as default.

Then why is 'use NNUE ' false by default?

I have compiled and tested on my little pi 4 and Rubichess 1.9 is stronger without the depth 16 own net.

[Madeleine Birchfield]

Andreas prefers NNUE as default.

Then why is 'use NNUE ' false by default?

Okay, there is some talk about latest Rubi here so I give some answers:

1. I'm not too proud about Rubi now being another NNUE player so I decided not to make an anouncement here.
2. I did a release on my github cause people already started to use some unofficial (and badly compiled) binaries spread on some file hosters for their ranking lists.
3. "Use NNUE" is set to false by default because of 1. and because I want to be able to test at OpenBench in the future and wasn't sure if I can set options in the test setup.
4. In general I will leave it to the users to use Rubi with internal evaluation or using the NNUE but Graham has set Rubi to play his Div1 and there even with adaption of the Cfish NNUE eval code (different to most other engines Rubi has a different piece representation and needs some adaption) and the usage of SF trainer to generate a network from own training data Rubi is probably still more original in Div1 than at least five contenders.

Regards, Andreas

Regarding 1. and 3. have you considered implementing your own NNUE inference code and trainer instead of relying on Stockfish's? I think there are a few engines using OpenBench for testing who have their own NNUE implementations.

[jshriver]

Kouri 1.15
https://github.com/hamzainaan/Kouri/releases/

Sugar-NN 150120
https://github.com/Zerbinati/SugaR-NN/releases

Anyone have a copy of these? They have been removed.

[tmokonen]

Kouri 1.15
https://github.com/hamzainaan/Kouri/releases/

Sugar-NN 150120
https://github.com/Zerbinati/SugaR-NN/releases

Anyone have a copy of these? They have been removed.

I have the source and executables for Kouri, and the executables for that version of SugaR. I did download the source for SugaR, but that file got corrupted. I uploaded them here:

https://mega.nz/folder/m1lHjKKR#KU7GCykpAB5lIp4N19aP-w

[Frank Quisinsky]

Hi there,

Topple 0.8.0 is available.
https://github.com/konsolas/ToppleChess/

Best
Frank

[bastiball]

Hi there,

Topple 0.8.0 is available.
https://github.com/konsolas/ToppleChess/

Best
Frank

Hi Frank,
The code execution cannot proceed because libwinpthread-1 was not found