Search found 3803 matches

by jdart
Fri Mar 27, 2020 2:49 am
Forum: Computer Chess Club: General Topics
Topic: Tactical mistake by Stockfish (Grischuk-Ding at candidates)
Replies: 21
Views: 2128

Re: Tactical mistake by Stockfish (Grischuk-Ding at candidates)

The whole line starting 12. Nxe5 and going through 15. .. a5 has actually been played in multiple games, for example Ivanchuk-So, Tata Steel-A 2015. Pavlovic annotating this says "it had been worked out by computers," so it was known previously and apparently found by at least some engines, maybe in...
by jdart
Wed Mar 25, 2020 3:03 pm
Forum: Computer Chess Club: General Topics
Topic: New Tool
Replies: 55
Views: 6964

Re: New Tool

I think the main problem with test sets like STS is age That's a valid point. Unless there's a forced mate or tablebase win, deeper search can always possibly find a hole in existing analysis. Although my understanding is STS is really designed to test positional understanding, not tactics, so that...
by jdart
Tue Mar 24, 2020 1:20 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Perft function too slow because of vcruntime140.dll
Replies: 25
Views: 1873

Re: Perft function too slow because of vcruntime140.dll

The general rule is, don't do dynamic memory allocation (on the heap) for anything. Especially not anything that might be time-critical.

--Jon
by jdart
Fri Mar 20, 2020 10:25 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: repackaging of Fishtest's SPRT calculation code
Replies: 0
Views: 891

repackaging of Fishtest's SPRT calculation code

I have created a separate repository containing just the "stats" directory from Fishtest, as a single Python 3 package. It is self-contained. See: https://github.com/jdart1/stats This implements the pentanomial SPRT model. The original code is from Michel Van den Bergh and I have only made small cha...
by jdart
Sun Mar 15, 2020 1:24 pm
Forum: Computer Chess Club: General Topics
Topic: Engine programming
Replies: 32
Views: 2879

Re: Engine programming

I agree, C++ is quite difficult for a beginning programmer, and if you are learning the language, a chess program probably shouldn't be your first project.

--Jon
by jdart
Sat Mar 14, 2020 6:47 pm
Forum: Computer Chess Club: General Topics
Topic: LC0 v0.24.0 release
Replies: 11
Views: 2161

Re: LC0 v0.24.0 release

Windows 7 (as well as earlier versions) is unsupported as of early this year. That was after 10 years of support. I am not a Microsoft fan, and they've done some bad things as a company, but having to upgrade your OS once every 10 years or so is pretty expectable. Staying on an unsupported OS is not...
by jdart
Mon Mar 09, 2020 7:07 pm
Forum: Computer Chess Club: General Topics
Topic: interesting endgame
Replies: 1
Views: 432

interesting endgame

From CEGT's 40/120 game database. Interesting how Fizbo has big + evals throughout the endgame. But Stockfish and Houdini think 58. .. Ne2+ draws. [Event "CEGT 40/120"] [Site "?"] [Date "2020.01.18"] [Round "16.1"] [White "Fizbo 2.0 x64"] [Black "RofChade 2.2 x64"] [Result "1-0"] [ECO "D85"] [PlyCou...
by jdart
Sun Mar 08, 2020 12:17 am
Forum: Computer Chess Club: General Topics
Topic: Is this English opening a win for white?
Replies: 8
Views: 1021

Re: Is this English opening a win for white?

After 1. c4 b6 2. Nc3 e6 3. e4, .. f5 as in your first game is unusual and probably not good, although Black plays f5 later in some lines. . .. c5 as in the second game would usually transpose to a Hedgehog (A34). Also possible is ... Bb7. Those options are not terrible for Black, in general. But I ...
by jdart
Thu Mar 05, 2020 10:56 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Important evaluation terms in this kind of position
Replies: 6
Views: 1221

Re: Important evaluation terms in this kind of position

Pawn attacks on the King zone are tricky. Sometimes that kind of thing succeeds. Sometimes it doesn't. However, I note current Stockfish seems to like that kind of attack a lot and it has a good success rate with it. Another factor here is that the Black Queen is on the wrong side of the Pawn chain ...
by jdart
Mon Mar 02, 2020 10:54 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Searching fail highs shallower..
Replies: 16
Views: 2230

Re: Searching fail highs shallower..

Note this technique is usually applied only at the root, although I suppose it could be done at interior nodes. I have implemented this in Arasan and it was an ELO gainer. I also tried limiting the amount of reduction--right now it reduces on ply for every fail high iteration--but that gave inconsis...