New Chess Engine: RukChess 3.0

Questions and comments related to CCRL testing study
Post Reply
IlyaRuk
Posts: 2
Joined: Fri Sep 09, 2022 4:55 pm
Sign-up code: 10159

New Chess Engine: RukChess 3.0

Post by IlyaRuk »

Hello CCRL community!

My name is Ilya Rukavishnikov and I would like to introduce the chess engine I have developed to the community. I can't say that it is something unique. It incorporated many different ideas from the many open source chess engines I studied (in my code I tried to indicate such engines as much as possible). I started working on it back in 1999 by studying and trying to run the code, which can now be found at https://www.ioccc.org/years.html#1992_vern. Then it became very interesting for me to figure out how a computer can play chess. I only had the source code for this program and the source code for GNU Chess, which, at the time, seemed too complicated to me. Then, I left the study of this topic for many years and returned to it only in 2016. Over the next 6 years, I read and studied the source code (special thanks to GitHub) of several dozen chess engines (from TSCP to Stockfish), many scientific articles and conducted hundreds (maybe even thousands) of experiments. I've got folders full of code printouts, notes, and charts. I still have an idea to turn this into an undergraduate course (I've been a university programmer for two decades and have teaching experience), but I'm afraid I won't be able to find the time for it. But I decided to start somewhere :) And the first thing I did was publish a few days ago the source code of my chess engine, executable files for Windows, a trained neural network, a modified trainer code based on berserk-trainer (https://github.com/jhonnold/berserk-trainer), a web version of the interface, and several utilities. All this can be found on my GitHub page (https://github.com/Ilya-Ruk). Also there are links to the source codes of programs, articles and the results of some of my tests. I would like this engine to be added to the rating list. In conclusion, I would like to say a huge thank you to the community for creating the rating list and this forum. All this had a huge impact on the development of computer chess in the world and encouraged many people to learn something new.

Best regards, Ilya Rukavishnikov.
Last edited by IlyaRuk on Sun Sep 11, 2022 8:33 pm, edited 1 time in total.
User avatar
Graham Banks
Posts: 26925
Joined: Sun Dec 18, 2005 5:47 pm
Sign-up code: 0
Location: Auckland, NZ

Re: New Chess Engine: RukChess 3.0

Post by Graham Banks »

How strong roughly, do you expect RukChess 3.0 to be in CCRL terms?
IlyaRuk
Posts: 2
Joined: Fri Sep 09, 2022 4:55 pm
Sign-up code: 10159

Re: New Chess Engine: RukChess 3.0

Post by IlyaRuk »

For RukChess 3.0 NNUE 2 with the rukchess.nnue network (available in the release https://github.com/Ilya-Ruk/RukChess/releases/tag/3.0) I ran 3 tournaments of 1000 games with each of the 3 engines (see below ) and calculated the average ELO of the program. The ELO of 3 engines was taken from the rating https://www.computerchess.org.uk/ccrl/404/. Details are available at https://github.com/Ilya-Ruk/RukChessNet ... ch-40-used.

ELO calculation

Hakkapeliitta 3.0 x64: 2984 + 195 = 3179
Demolito 2021-10-04: 3166 - 7 = 3159
Xiphos 0.6 BMI2: 3332 - 164 = 3168

Avg. 3169 ELO
User avatar
Graham Banks
Posts: 26925
Joined: Sun Dec 18, 2005 5:47 pm
Sign-up code: 0
Location: Auckland, NZ

Re: New Chess Engine: RukChess 3.0

Post by Graham Banks »

Thanks. :thumbup:
User avatar
Graham Banks
Posts: 26925
Joined: Sun Dec 18, 2005 5:47 pm
Sign-up code: 0
Location: Auckland, NZ

Re: New Chess Engine: RukChess 3.0

Post by Graham Banks »

I think that RukChess 3.0 is too strong for my Amateur Division 5, so I'll make sure to include it in the next series.
Post Reply