CCRL Discussion Board

I've just replied to your mail, then noticed that it was already too late! So quoting my reply here. But basically it's good idea to ask in the forum first whenever possible. Dear Deterdettol, Thanks for your question and interest, and welcome to the forum! Your questions are most certainly appropri...

Your second method is roughly what I use in my solvers, it's called the "grandfather method". Is is described a bit here: https://chessprogramming.wikispaces.com ... #Algorithm and also in 1986 Ken Thompson's paper.

Hi Dan, Here is a standard technique for finding bugs in a tablebase. Iterate through all positions in a finished tablebase. For each position, make all possible moves, including captures and promotions (search 1 ply deep). Compute the position value based on this search, and compare with the stored...

Well the only way for us to provide such deltas is by running diff on complete databases of two weeks. Simply uploading a separate PGN of only last week's games is not enough, because sometimes engines are removed or renamed, or errors are corrected, which affects older games. We take special care t...

Hello everyone, I would be grateful if you could give me answers to my questions: 1. Can I use only vertical/horizontal symmetry (i.e. restricting king to the a1-d1-d4-a4 square) in the pawnless endgame? I ask this question, because in early work they allowed black king to stay in the a1-h1-e4-d4 t...

Hi Vincent,

If I remember correctly, Wilhelm should be able to do this sort of mining. Have you tried it?

Doirse wrote:(krkp - 4,054,640 ESM)

krkp - 4,054,640 positions with white to move, 4,986,658 positions with black to move (listed as kpkr in NULP stats file).

Welcome and good luck in your search!

this is incredibly fascinating! It is! Is it possible to create W/D/L results for all possible positions in a specific endgame? For example, looking at the NULP results for KRKP endgames there are 9,041,298 possible legal positions. How would one go about determining how many of those are W/D/L for...

Someone asked me by email: Hello Kirill, Could you please explain me how should I interoperate the analysis results of the Syzygy EGTB running on Houdini 4. I am attaching a screenshot and would like to understand the meaning of the numbers within the parentheses. Thank you, Syzygy-egtb-question-ima...

In which direction would you say the error for top engines is? Carlsen is currently rated FIDE 2863. Stockfish is currently rated CCRL 3314, so theres a 451 point difference. Do you thinks this 451 over or under estimates the real performance difference? In other words: A: Are top engines even stro...

1. Does anybody know if Stockfish plays better/worse on PC vs. Mac (with equivalent settings)? Is there any reason to think that chess engines running through Mac OS might be inferior? Superior? --> For instance would Stockfish for Mac out play Stockfish for PC or vice versa? If there's a discernib...

Perhaps a larger hash is dictated for the 10-15 minute analysis, a medium hash for 2 minute analysis? Yes, but I think the difference is not significant enough to worry about. 512 MB for 2 minutes and 2 GB for 10-15 minutes should be more than enough, IMO. I don't think it makes sense to keep chang...

In general hash size has very minor effect on engine performance (outside of extremely small sizes). The gain from hashtable size saturates at some point, depending on time control and hardware speed. Any engine playing blitz game won't benefit much from a 64 GB hashtable, at least not measurably mo...

Welcome and good luck with this project! Elo weights recent games more heavily than older games (weighted average). True for human games, but not in computer chess. In engine games generally all games are weighted equally when computing the ratings. 1. Where can I find the data in a manageable form ...

I understand that you run BayesElo with the whole database before each list update to get the ratings, am I right? Yes. If so, how do you precisely then determine the number to add to the ratings to get the numbers given in the list? Reason for asking: I'd like to know roughly the "CCRL rating...

Hi and welcome back! This is very much on topic here. Unfortunately I have no answer, as I never used RetroChess. Hopefully someone can help.

(Also, If I may, I'll point out that this question demonstrates how closed probing code may be a major obstacle to tablebase format adoption).

Here is one relevant chart: http://rwbc-chess.de/Additional/chronology.htm I think there should be more than 1000, just counting engines. Much larger number if you also counted non-engine chess program, including those for all obscure and ancient devices. (Also, it will depend a lot on your definiti...

Someone asked me about the number of positions with en passant capture rights, and how it compares to number of positions with castling rights. So I added such numbers to the NULP site ( link ). In particular, the following chart may be interesting. http://kirill-kryukov.com/chess/nulp/images/Propor...

Can you tell me how to access the message recently posted by me in the Endgame Tablebases forum? Endgame Tablebases is a public forum, so your post should be accessible by you as well as anyone else by following this link: http://kirill-kryukov.com/chess/discussion-board/viewtopic.php?f=6&t=743...

The number of 8-piece FRC positions: 39,587,704,309,395,475.

Hi galen, thank you for sharing your very interesting results!

Can you verify your tables against Kronsteen's ones?

Do you think you could compute all 6-piece tables using 64 GB or RAM?

Are you planning to release the generator, the tables, or to make a web-interface?

Welcome to this forum!

The number of unique legal positions in 8-piece chess endgames: 38,176,306,877,748,245. Not sure if I'll count for 9-pieces.

Here is a Windows 64-bit binary of Stockfish 4 with added syzygybases access capability. Compiled with MinGW GCC 4.8.1 from Cygwin64 distribution. Non-PGO build, "x86-64-modern" architecture (with popcnt). The source was obtained from here (link was posted by Ronald on CCC).

Yes, this archaic software unfortunately has some issues with Vista and 7. I managed to install it on Vista few years ago, but I forgot the details now. Still, it's excellent once you manage to install it. May be try experimenting with compatibility options, or install XP in a virtual machine. Sorry...

Hello muzzoid, and welcome to this forum! About the Lomonosov 7-man tablebases - they may be complete, but they are not available (unless I am missing something), as well as the generator they used. So there is still work to be done in this area. In 6-man tablebases, probably syzygy's project is the...