Search found 3349 matches

by Sven
Sun Sep 20, 2020 5:01 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: PLEASE HELP with MATING scores adjustments in TT
Replies: 9
Views: 539

Re: PLEASE HELP with MATING scores adjustments in TT

In your search, the mate distance is relative to the root position. If you store it like that in the hash tables, and you have a hit in the next move turn, the mate score is wrong. Example. You are at move 20 and see a mate in 8 plies. You store that position. Next move turn, so you are at move 21,...
by Sven
Mon Sep 14, 2020 9:13 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: A very quick clarification question on sorting PV moves
Replies: 5
Views: 298

Re: A very quick clarification question on sorting PV moves

[...] assuming that what you call parent node is what I called previous ply's move (it's literally the same right? say: e2e4 e7e5 - here if we are at e7e5 then e2e4 would be parent root node or how I call it previous ply's move) IS actually "leading to the current node" or how I call it - is the mo...
by Sven
Sun Sep 13, 2020 10:03 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: A very quick clarification question on sorting PV moves
Replies: 5
Views: 298

Re: A very quick clarification question on sorting PV moves

It is as simple as that: 1) A node belongs to the root PV (i.e., follows the root PV) if its parent node belongs to it and the move leading to the current node is the parent's PV move. 2) The root node always belongs to the root PV. In an older private engine I developed years ago I implemented that...
by Sven
Thu Sep 10, 2020 12:09 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Speeding Up The Tuner
Replies: 38
Views: 1587

Re: Speeding Up The Tuner

Even when your evaluator was piece-square only it couldn't be much faster than reading a FEN. But if you clear a few GB of memory that was not used in the first place... That is just wrong design. There is no reason at all to clear the hash table on setboard. There actually is never any need to cle...
by Sven
Tue Sep 08, 2020 10:22 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: I don't want to give up
Replies: 47
Views: 2292

Re: I don't want to give up

Hi Sven, The legality check does more than legality checking. It scores each move. The idea is since this is done at the leaves anyway doing it all the time does not cost much and gives the added benefit of good move ordering. But if I am wrong about this untested assumption it can be changed later...
by Sven
Tue Sep 08, 2020 10:14 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: I don't want to give up
Replies: 47
Views: 2292

Re: I don't want to give up

As to the speed, see my EDIT in my previous post above ... Changing the move generator from m->score = -Qsearch(t, m + 1, -beta, -alpha); into something like this: s32 KingInCheck(Thread * t, s32 side) { return (side == BLACK) ? AtkByWhite(t, one << king[BLACK]) : AtkByBlack(t, one << king[WHITE]); ...
by Sven
Tue Sep 08, 2020 9:26 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: I don't want to give up
Replies: 47
Views: 2292

Re: I don't want to give up

Well I thought i had found a bug. In Search() sna (score, move-count, alpha) was being returned without including n from snu. But that change changed nothing as far as I could tell. Also I created the Snu union so I could return a u64 value with the score and the move count. Instead I was returning...
by Sven
Tue Sep 08, 2020 10:56 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: I don't want to give up
Replies: 47
Views: 2292

Re: I don't want to give up

Well I thought i had found a bug. In Search() sna (score, move-count, alpha) was being returned without including n from snu. But that change changed nothing as far as I could tell. Also I created the Snu union so I could return a u64 value with the score and the move count. Instead I was returning...
by Sven
Sat Sep 05, 2020 8:08 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: Mr Bob And The Raven
Replies: 9
Views: 1634

Re: Mr Bob And The Raven

Thanks Graham. It does not happen too often that Jumbo wins a tourney so I am glad about this result! :wink:

This is still the old Jumbo from last year, and I also did not produce anything new in the meantime due to lack of spare time ... Don't know when this will change - we'll see.
by Sven
Sat Aug 08, 2020 12:56 pm
Forum: Computer Chess Club: General Topics
Topic: this will be the merge of a lifetime : SF 80 Elo+
Replies: 121
Views: 14110

Re: When to check NNUE box

"The idea is to use NNUE only on quite balanced material positions." This was based on the fact that SF-NNUE searches much more slowly than classic SF on most hardware. But if one has fast hardware, so that SF-NNUE runs fast enough, then why not always use SF-NNUE? How is stockfish defining 'quite ...