New engine: CederChess

[Claus Cederberg]

Thanks Gabor,

I have mentioned 1400 ELO, but I can see on the CCRL Blitz list that the ELO scores are lower than they was a couple of months ago - at least for the weaker engines.

Based on many games against the engines Ozwald, Phalp and Quokka, I estimate CederChess 2 to have an ELO in the 1250 range on this list.

Best regards,
Claus

[Gabor Szots]

Thanks Gabor,

I have mentioned 1400 ELO, but I can see on the CCRL Blitz list that the ELO scores are lower than they was a couple of months ago - at least for the weaker engines.

Based on many games against the engines Ozwald, Phalp and Quokka, I estimate CederChess 2 to have an ELO in the 1250 range on this list.

Best regards,
Claus

Tournament running. It won't be completed till this week's update, though.

The method of calculating the Elo ratings for our lists has changed recently, that's why the values changed. The intention is to make the error margins sensible.

The engines you used are also participating in my tournament. Early results seem to confirm your estimation.

Best regards,
Gabor

[Claus Cederberg]

Thanks for adding CederChess 2 to the list, Gabor

I do have some improvements in mind with the current architecture. I will work with these next winter.

To make a huge jump, I think a new architecture is needed.

[Gabor Szots]

Second run of CederChess 2 64-bit in progress.

Watching some games I noticed that it moves too quickly, sometimes almost instantly. It has more time at the end of the game than at the start. The time control I use is 1:30+1 (90+1).

[Claus Cederberg]

I may have underestimated your hardware, Gabor

According to https://computerchess.org.uk/ccrl/404/about.html you use a Intel Core i7-4770K, while I have a Intel Core i7-1165G7 in my computer. So I made this processor comparison: https://www.cpubenchmark.net/compare/19 ... -i7-1165G7

The processor you use is around 30% slower than mine, so my own tournaments have a 60/0,7 time control. I'm also a little conservative, so I want to be completely sure, CederChess-2 don't lose on time.

But it is great news that it moves so quickly on your hardware. That means I can replace the current very simple evaluator with a more advanced evaluator. I have already written the advanced evaluator, but I did not include it, because I was afraid of the time usage.

I will not provide a new version with the advanced evaluator now. I have some other ideas to also try out, to make sure the ELO gain on the next version will be significant.

[Gabor Szots]

Hi Claus,

My CPU is an i7-12700K. The i7-4770K is our reference machine. We run the Stockfish 10 bench command to determine how our machine compares to the reference machine, that's why I use 90+1 instead of 120+1.

Anyway, I think the time management of an engine should suit all machine speeds and time controls, regardless of how complicated its evaluation is.

Gabor

[Claus Cederberg]

In the next version I'll implement better time management.