TalkChess.com

Hi all!

I just got into the hobby of chess engines, and I am finally happy enough with the start of this engine to release it. It is heavily based on Vice by bluefever, but it should play around 2200 ELO. It is written in C, and is UCI compliant.

the GitHub link: https://github.com/bctboi23/seeChess

Any feedback you have is much appreciated!

bctboi23 wrote: ↑Fri Feb 07, 2020 6:33 pm Hi all!

I just got into the hobby of chess engines, and I am finally happy enough with the start of this engine to release it. It is heavily based on Vice by bluefever, but it should play around 2200 ELO. It is written in C, and is UCI compliant.

the GitHub link: https://github.com/bctboi23/seeChess

Any feedback you have is much appreciated!

Um.....I think there was/is already a engine named See Chess...v0.6.9....here is a copy...not a bad chess engine..

https://www.4shared.com/s/ferplqpqPea

supersharp77 wrote: ↑Fri Feb 07, 2020 8:28 pm

bctboi23 wrote: ↑Fri Feb 07, 2020 6:33 pm Hi all!

I just got into the hobby of chess engines, and I am finally happy enough with the start of this engine to release it. It is heavily based on Vice by bluefever, but it should play around 2200 ELO. It is written in C, and is UCI compliant.

the GitHub link: https://github.com/bctboi23/seeChess

Any feedback you have is much appreciated!

Um.....I think there was/is already a engine named See Chess...v0.6.9....here is a copy...not a bad chess engine..

https://www.4shared.com/s/ferplqpqPea

Gosh dangit! I guess great minds think alike

I didn't see an engine with this name anywhere (not on the CCRL either). @supersharp77 Is there another list of engines somewhere?

The old version of SEE is on the CCRL rating list. It is just called SEE 0.6.9 there, and is only on the blitz rating list, not the 40/15 list. Technically, your engine name is slightly different, because see is all lower case, and there is no space in the name of your engine.

bctboi23 wrote: ↑Fri Feb 07, 2020 6:33 pm Hi all!

I just got into the hobby of chess engines, and I am finally happy enough with the start of this engine to release it. It is heavily based on Vice by bluefever, but it should play around 2200 ELO. It is written in C, and is UCI compliant.

the GitHub link: https://github.com/bctboi23/seeChess

Any feedback you have is much appreciated!

I have added your engine to the XB/UCI chronology. (see link in my signature)

Currently I took your first github release from 2020-01-03 (version 1.0) as first release, because your github tells so.
Ofc I can change this, if you like that your announced version should be named as first official release.
[Collectors type of people will still dig out all versions available, if they find them ;-)]

Guenther wrote: ↑Sat Feb 08, 2020 9:41 am

bctboi23 wrote: ↑Fri Feb 07, 2020 6:33 pm Hi all!

I just got into the hobby of chess engines, and I am finally happy enough with the start of this engine to release it. It is heavily based on Vice by bluefever, but it should play around 2200 ELO. It is written in C, and is UCI compliant.

the GitHub link: https://github.com/bctboi23/seeChess

Any feedback you have is much appreciated!

I have added your engine to the XB/UCI chronology. (see link in my signature)

Currently I took your first github release from 2020-01-03 (version 1.0) as first release, because your github tells so.
Ofc I can change this, if you like that your announced version should be named as first official release.
[Collectors type of people will still dig out all versions available, if they find them ]

Great, thanks!

I would prefer to have v1.2 looked at as the first official release. I delayed creating an account and posting this engine until now because I thought it was still a practical copy of Vice until v1.2, so in my opinion, v1.2 is the first release because it is the first time that seeChess is different enough to be considered a different engine. Of course, a LOT of code is still from Vice, and I hope to be changing that very soon (next project is overhauling the evaluation function).

Friendly advice, don't start with the evaluation function.

The real magic is in search. A simplistic eval with half-decent values will do the job as you improve the search code, and clean other parts like movegen to be more efficient.

Even a very good eval would still net you a weak engine if the search and other parts aren't up to par.

You should also use SPRT to validate your changes, don't just rely on how you feel about a change or limited (and thus unreliable) tests.

Another periphery question:

Is there some sort of procedure to get an engine on a rating list (like CCRL)? Do I have to submit the source to another forum or something?

@Alayan, I will definitely keep that in mind! The plan right now is to add some better piece square tables and a tapered eval, and then work on pruning even more. Right now, I feel the evaluation is too bare bones to do any more pruning than I have already added.

Also, what is SPRT? I have been doing tests on my machine (1000 games against a couple engines around the 2200 level), is SPRT a better way to test?

bctboi23 wrote: ↑Sun Feb 09, 2020 5:30 am Another periphery question:

Is there some sort of procedure to get an engine on a rating list (like CCRL)? Do I have to submit the source to another forum or something?

All you have to do is announce that you have a release version of your engine here. Most of the CCRL testers come here regularly.

bctboi23 wrote: ↑Sun Feb 09, 2020 5:30 am @Alayan, I will definitely keep that in mind! The plan right now is to add some better piece square tables and a tapered eval, and then work on pruning even more. Right now, I feel the evaluation is too bare bones to do any more pruning than I have already added.

Tapered eval is useful and easy to add. It gave me a nice Elo boost when I added it to my engine.

bctboi23 wrote: ↑Sun Feb 09, 2020 5:30 am Also, what is SPRT? I have been doing tests on my machine (1000 games against a couple engines around the 2200 level), is SPRT a better way to test?

Take a look at https://www.chessprogramming.org/Match_Statistics for a description.

To remove the ambiguity of naming (SEE is a little to close to See imo) I renamed the engine to "CeeChess"

I added the tapered eval with some new piece square tables (Ludmyil's PSQTs are great!) and got an extra 100 ELO selfplay.

@Guenther, would you be able to change the name in the spreadsheet? Sorry for the inconvenience, I just feel like it should fix the name issue