TalkChess.com

After some sleep I realize I wrote a bunch of nonsense. :D 2^188 < 768 choose 32 < 2^189, gives a lower bound for 189 Bits. should be correct though I think. Simply by starting from an empty board, all possible positions are a subset of all 32 key combinations. (plus less pieces but that should be n...

Unfortunately Hufman encoding cannot be updated incrementally in an efficient way. A Zobrist approach with a very long key is much more practical. Human encoding of a position with all pieces present would take 32*1+16*3+12*5+4*6 = 164 bits (where castling rights can be kludged in by representing v...

This made me think, what if you use a collision free board hashing + lockless hashing to possibly (?) get rid of these problems. Here's some thoughts on that http://talkchess.com/forum3/viewtopic.php?f=7&t=72964 I don't think this gains Elo but it might at least be interesting to think about and/or ...

So I was wondering how expensive it would be to get a collision free hash code for positions so that in search one doesn't have to worry about collisions as well as it possibly helping for perfting where a hash collision really makes a difference. It's trivial to get 256 Bits and user Pio had the ni...

If you're looking to run very long searches, the standard Leela binary will not be very useful since AFAIK they haven't changed the data type of their Q. (at least last time I checked Q was a float which means that for long searches (> 100MN) the eval and with it the search will become stuck due to...

Which network will Lc0 be using? Default or specialized settings? Also, what endgame tables will be available to your engines? Sergio 384x30-t40-1808 with specialized settings: in-between default and "Kiudee." If you have any suggestions about settings, please let me know. I have 6-men tablebases b...

I haven't implemented multi PV in my engine so I might be completely wrong but couldn't one reason why it is so slow be that the TT gets trashed from all the different bounds? In single PV mode if the eval is somewhat stable all bounds for all positions will be around that eval, either as upper or l...

I had planned to send this version of my engine to CCRL testing but after this disastrous tournament I may have to write a few improvements first.

Sergio Martinez wrote: ↑

Fri Jan 10, 2020 9:52 pm

koedem wrote: ↑

Fri Jan 10, 2020 1:59 pm

Should one be able see live games? I can just see an empty chess board when following the link.

Yes, it is exactly Alexander says.

Strange, I kept checking every few hours but never saw anything.

Should one be able see live games? I can just see an empty chess board when following the link. Also seems like for my engine nothing changed, still getting good positions and blowing them. :D I wonder if in this field where my engine is weakest it might be a good idea to set a high draw score, that...