My test for Kayra 1.0 v6 bmi2

[AlexChess]

Dear Eduard,

I have posted a crosstable and some games of Kayra 1.0 & 1.1 against latest SF compiled by Archimedes and SF+ compiled by Firefather in my Android engines tournaments section on TC. Hoping that it will be useful to you and Mehmet for compiling that Kayra.bin to improve the strength of Kayra 1.1 even more

forum3/viewtopic.php?f=6&t=78803

Best regards, Alex

[Eduard]

I started playing with Kayra.bin on Playchess. So far the book has played well, not yet defeated. I can only improve the book where it has been lost. And of course where they won.

So far I've collected a few winning games and a friend also plays with Kayra 1.1 and 64 cores on Playchess and he keeps winning! All of this comes in the book update that I'm planning for January 6, 2022.

[AlexChess]

NB: I'm not saying copying swathes of SF (etc) and leaving it unchanged necessarily violates the licence, rather that is it boring, intellectually void(), not satisfying and not creative etc. To do so just for elo is lacking integrity. Until very recently I also assumed that it totally wastes the time of enthusiasts who spend their valuable cpu hours testing engines and having fun and posting useful results back to developers; but on this I seem to have been impressively wrong.

It's just a thought but maybe it would be useful to have a register of testers like Ed's Programmer Code of Honour. If registered you guaranteed (within reason) not to post results from engines that violate a licence (or the posting violates a licence) and (tricker) not to post results from intellectually void engines - e.g. a clone of SF with some constants changed. Maybe if wording could be agreed Ed could be persuaded to add a page to his website as he has the necessary gravitas.

http://rebel13.nl/download/programmer-c ... honor.html

I've followed your suggestion, I've removed all not GPLv3 compliant engines that I was testing in my SuperBlitz and I will respect Ed's Programmer Code of Honour when I will register and modify SF source, publishing obvioulsy also the modified source code. Browsing the Rebel site, I have discovered that there is also a RebelFish and testing it right now! https://rebel13.nl/home/rebelfish.html

[op12no2]

NB: I'm not saying copying swathes of SF (etc) and leaving it unchanged necessarily violates the licence, rather that is it boring, intellectually void(), not satisfying and not creative etc. To do so just for elo is lacking integrity. Until very recently I also assumed that it totally wastes the time of enthusiasts who spend their valuable cpu hours testing engines and having fun and posting useful results back to developers; but on this I seem to have been impressively wrong.

It's just a thought but maybe it would be useful to have a register of testers like Ed's Programmer Code of Honour. If registered you guaranteed (within reason) not to post results from engines that violate a licence (or the posting violates a licence) and (tricker) not to post results from intellectually void engines - e.g. a clone of SF with some constants changed. Maybe if wording could be agreed Ed could be persuaded to add a page to his website as he has the necessary gravitas.

http://rebel13.nl/download/programmer-c ... honor.html

I've followed your suggestion, I've removed all not GPLv3 compliant engines that I was testing in my SuperBlitz and I will respect Ed's Programmer Code of Honour when I will register and modify SF source, publishing obvioulsy also the modified source code. Browsing the Rebel site, I have discovered that there is also a RebelFish and testing it right now! https://rebel13.nl/home/rebelfish.html

I wish you well Alex. I don't pretend to understand the fascination with engines like Kayra (no stated creative goals) but that's a limitation of mine I guess. Have fun.

[Eduard]

Kayra.bin:
I started my test on Playchess and after 70 games there is still no defeat. The problem: Kayra plays very well and on Ryzen 3900X Kayra cannot be defeated even with the little Kayra.bin.

Alternatively, I will therefore deliberately throttle the hardware to see what it looks like with Kayra.bin. For that I need more time because I want to play 300 games. Currently is played with 22 threads, after 300 games I only play with 2 threads (1 core).

[Eduard]

A few words about Kayra.bin:

Draw variants (important for slow hardware) are implemented enough.

Killer variants? Yes, I have implemented all known killer variants that are known on PlayChess.

Winning variants? Yes, there are many in the book.

Which variants are still missing? There is no point in importing all games. Many such games are decided by simple mistakes or by lack of time. Therefore, only winning variants make sense, the initiative promise after the last book move! These are then often a few new variants. A BiG book (with all possible games) is not better than a small book made up of the best variants.

So far, 700 games that Kayra has played have also been implemented.

[Sylwy]

A few words about Kayra.bin:

Draw variants (important for slow hardware) are implemented enough.

Killer variants? Yes, I have implemented all known killer variants that are known on PlayChess.

Winning variants? Yes, there are many in the book.

Which variants are still missing? There is no point in importing all games. Many such games are decided by simple mistakes or by lack of time. Therefore, only winning variants make sense, the initiative promise after the last book move! These are then often a few new variants. A BiG book (with all possible games) is not better than a small book made up of the best variants.

So far, 700 games that Kayra has played have also been implemented.

In my topic - Waterloo - in this section, Kayra PB 1.1 plays linked to Kayra.bin book (like engine book). A very good one.

[Eduard]

Oh, thank you, great. I play live on playchess and after 114 games I am still undefeated. I watch a lot of games and have discovered some things that can be improved. Only the second best move was played in a game; the best would have won. So I can say that with the upcoming book update (next week) the book will play better. Could you post the losing games in PGN in your Woterloo thread?

Thank you.

[AlexChess]

Oh, thank you, great. I play live on playchess and after 114 games I am still undefeated. I watch a lot of games and have discovered some things that can be improved. Only the second best move was played in a game; the best would have won. So I can say that with the upcoming book update (next week) the book will play better. Could you post the losing games in PGN in your Woterloo thread?

Thank you.

Yes, but Kayra never loses...

[Sylwy]

Could you post the losing games in PGN in your Woterloo thread?

Done.