TalkChess.com

If you want to try many engines and/or make matches between them, download them all. Otherwise, syzygy is the best in many aspects, 5 men is the easiest one to get.

Take a look their brief information:

https://www.chessprogramming.org/Endgame_Tablebases

OK, sorry, my mistake. It was not my intention that the black Cannon would ever be protected. So imagine the Elephant on e7 is not there at all. From your second posting I understand that for the 1.Hd4, where the Horse chases the Cannon and at the same time the Cannon offers exchange, you would not...

That is why my program rules games in two separating steps, the first one is to detect continuously chasing without taking care the relationship between attackers - victims, in this case the Horse is doing that but not that black Cannon since it is not continuously chasing the white Cannon (in one ...

I am in doubt about the priority of the rule that it is allowed to perpetually offer exchange of equal pieces. One way to look at it is that the possibility to exchange is as good as being protected, i.e. the piece is capable of self-defence, and initiating the exchange is like a 'pre-emptive' reca...

So it really was a serious question: how long do you take to generate, say, KBNK? Sorry I don’t know. I don’t generate EGTB for chess but for other chess variants. ... Because chasing can heave multiple targets (which then is allowed) Technically speaking, chasing multi-pieces is a draw. You can ig...

Well, if you would require several hundred times as much calculation per position as is really needed, I would have no difficulty believing that. So the key question is: how fast is your generator (on a single-threaded CPU)? I think you know already, there is no magic in computer chess. Except the ...

Another question would be: "why do it at all"? For orthodox Chess all 7-men have already been done. And 8-men are too big to store the result. A reason for doing it on a GPU would be for fun. I think that the GPU would be a very natural place for a mate solver that uses proof search. That is becaus...

From what I recall, tablebase generation is generally I/O bound, not CPU bound. In the EGT generator on my old website (basically FairyGen) profiling showed that the single statement data=EGT[index] was responsible for 99.5% of the execution time. The entire remainder of the calculation only took 0...

Thanks a lot for useful information! I have been reading about accessing RAM from graphics cards. Just read that AMD has been targeting on that issue for a while. However I am not sure if they are successful (accessing randomly RAM with low latency from GPU). Does anyone have information/confirmatio...

We have been working on EGTB generators for a while. One of the main problems beside the need of huge memory, is that those generators are always hungry for computing power, more CPUs/cores is better. One of idea we want to try is to use graphics cards for generating. AFAIK, a good graphics cards ma...