Chenglite (UCI) by Maksim Korzh

Norbert Raimund Leisner · Tue Sep 11, 2018 12:32 pm

Hello,

cf. successor of Chengine here: https://github.com/maksimKorzh/Chenglite - Linux, but not Windows binaries are included

Best wishes,
Norbert

tmokonen · Post by **tmokonen** » Tue Sep 11, 2018 8:21 pm

Version 1.1 is available for download on the GitHub site. I made a 64 bit Windows compile and placed it here:

http://tonyschess.x10host.com/Chenglite11TM.rar

CMCanavessi · Post by **CMCanavessi** » Wed Sep 12, 2018 5:41 pm

Do any of you know aproximate strength?

tmokonen · Post by **tmokonen** » Wed Sep 12, 2018 9:46 pm

It only does fixed depth searches for now, so I am not sure it would be appropriate for your tournament yet.

menniepals · Post by **menniepals** » Tue Sep 18, 2018 6:21 pm

CMCanavessi wrote: ↑Wed Sep 12, 2018 5:41 pm Do any of you know aproximate strength?

2100

Dariusz · Post by **Dariusz** » Tue Sep 18, 2018 6:57 pm

I made executable for macOS

https://www.solidfiles.com/v/WdWMDxNADmDLw

maksimKorzh · Post by **maksimKorzh** » Sat Sep 22, 2018 8:37 am

Hi guys! I'm really surprised to see the topic dedicated to my engine, thank you for your interest! Now I'll try to answer to all of your posts... so:

1. version 1.1 - forget about it due to it was full of bugs in search, kindly pointed out by Marco Belli and H.G.Muller, now the only version is available - version 1.0 which seems to be more or less stable... It is heavily improved compare to old 1.0 - iterative deepening and move ordering were added as well as quiescence search in no longer fixed to depth 4 + minor bug fixes, so you might consider current 1.0 version as 1.2

2. answer to yellow-panther who issued Chenglite on github: I'll make win x64 version in a couple of minutes and add another release file.

3. I can't make MAC release, so please recompile the current fixed source and give me a link - I'll add it as well.

4.

Re: Chenglite (UCI) by Maksim Korzh
Post by CMCanavessi » 12 Sep 2018, 18:41

Do any of you know aproximate strength?

Chenglite is not rated, it played only with Oscar Toledo's winboard tiny/nano chess and beats it easily. I appreciate if someone kindly takes Chenglite to the tournament and reports me the rating it'll gain. I believe it won't be higher than 1400 Elo.

5. someone asked about time controls: time management is not implemented. Chenglite can play fixed depth search games. If you set time control to blitz 5 + 0 or any other time related control like time per 40 moves or something - in all of these cases Chenglite will play at fixed depth 5.

6. Known issues:
- Chenglite will draw any winning endgame like King and Queen vs King for king evaluation for endgames is not implemented.
- Chenglite doesn't recognize 3 fold repetisions and 50 move rule.

Summury:

I'm aiming to get the minimalist chess engine like micro-Max by H.G.Muller. I've started to rewrite my engine from scratch (for the third time this month:) ) and now the design is completely different. This is going to be one search() function including inlined move generator and so on - just like in micro-Max. All the weak sides of Chenglite would be improved in my new engine Minilite, you can follow the development process here: https://github.com/maksimKorzh/Minilite This is a true minimalist approach which is my aim.

As for Chenglite - I feel completely satisfied with it that's why the development is finished. I hope it would be developed by the beginner community members. I don't mean the code to be that clean and clear, but in general it seems to be easier to understand compare to vice's or tscp's code.

Also I have a question to the community:
I'm thinking about whether to create a youtube tutorial, describing my new engine's design or not. Basically this it the micro-Max's design, described on H.G.Muller's site. Personally I've started to learn how to write a chess engine via typing from screen the code of vice by bluefever software, renaming the variables to get it clear how it actually works, so I wonder if someone else did something the same? And is it going to be interesting for anybody to type code from screen and to get micro-Max like engine? Waiting for your feedback, thanks.

maksimKorzh · Post by **maksimKorzh** » Sat Sep 22, 2018 8:59 am

windows version of Chenglite: https://github.com/maksimKorzh/Chenglit ... s/tag/v1.0

tpoppins · Post by **tpoppins** » Sat Sep 22, 2018 9:05 am

maksimKorzh wrote: ↑Sat Sep 22, 2018 8:37 am 5. someone asked about time controls: time management is not implemented. Chenglite can play fixed depth search games. If you set time control to blitz 5 + 0 or any other time related control like time per 40 moves or something - in all of these cases Chenglite will play at fixed depth 5.

What happens if it fails to reach d=5 before the time control is reached? Does it cut its search short or simply loses on time?

6. Known issues:
- Chenglite will draw any winning endgame like King and Queen vs King for king evaluation for endgames is not implemented.
- Chenglite doesn't recognize 3 fold repetitions and 50 move rule.

An engine that cannot mate a bare king with a queen? Can you sincerely claim that such an engine can play chess?

I'm aiming to get the minimalist chess engine like micro-Max by H.G.Muller.

Well, your goal is still a long way off, I'm afraid. micro-Max can play a reasonable game, observes all the rules and plays the KRK ending practically flawlessly.

maksimKorzh · Post by **maksimKorzh** » Sat Sep 22, 2018 9:30 am

tpoppins wrote: ↑Sat Sep 22, 2018 9:05 am
maksimKorzh wrote: ↑Sat Sep 22, 2018 8:37 am 5. someone asked about time controls: time management is not implemented. Chenglite can play fixed depth search games. If you set time control to blitz 5 + 0 or any other time related control like time per 40 moves or something - in all of these cases Chenglite will play at fixed depth 5.
What happens if it fails to reach d=5 before the time control is reached? Does it cut its search short or simply loses on time?

6. Known issues:
- Chenglite will draw any winning endgame like King and Queen vs King for king evaluation for endgames is not implemented.
- Chenglite doesn't recognize 3 fold repetitions and 50 move rule.
An engine that cannot mate a bare king with a queen? Can you sincerely claim that such an engine can play chess?

I'm aiming to get the minimalist chess engine like micro-Max by H.G.Muller.

Well, your goal is still a long way off, I'm afraid. micro-Max can play a reasonable game, observes all the rules and plays the KRK ending practically flawlessly.

1. It doesn't cut the search, so it'll lose on time
2. I don't claim anything. I really wonder about Chenglite's possible rating but I'm agree with you - it's not an engine for competitions, and won't be ever for I'm rewriting it now. I read about CCRL's demands and doubt Chenglite to pass even 50%.
3. My goal is definitely long way off, you're right, I'm a beginner and self learner. Nobody taught me to write chess engines and still my program can mate another one(the only one for now but still).

So I'm sorry if my request to rate my engine is a bit too early, I didn't mean to insult anybody.

One little question to you - what kind of software do you use in CCRL to compete engines? Can I try it myself at home to see, say whether my engine crashes or not and things like that, thanks in advance.