Strange Stockfish behavior?

[Uly]

And what output do you get on the GUI, does it show a5 failing high at depth 18 at second 32?

[Jeremy Bernstein]

And what output do you get on the GUI, does it show a5 failing high at depth 18 at second 32?

No, there is normally no output from the engine at the fail high/fail low, so no GUI output. I just wanted to show you what's happening internally.

Jeremy

[Uly]

Oh, thanks. So my request is to allow optionally that Stockfish sends those fail high/fail lows to the GUI (an option to not hide fail highs) sometimes that information is critical to find the best move, specially if I can find another problematic position in where Stockfish abandons the best move (that fails high internally).

Reporting fail lows is also very important, otherwise the user has to continuously check the move being analyzed to be aware that the main move could change. Currently the main move just disappears from analysis which is confusing (ideally, Stockfish would resolve the move that is failing low before looking for alternatives), it seems they actually managed to make Stockfish less user-friendly

[Jeremy Bernstein]

Oh, thanks. So my request is to allow optionally that Stockfish sends those fail high/fail lows to the GUI (an option to not hide fail highs) sometimes that information is critical to find the best move, specially if I can find another problematic position in where Stockfish abandons the best move (that fails high internally).

Reporting fail lows is also very important, otherwise the user has to continuously check the move being analyzed to be aware that the main move could change. Currently the main move just disappears from analysis which is confusing (ideally, Stockfish would resolve the move that is failing low before looking for alternatives), it seems they actually managed to make Stockfish less user-friendly

Your wish is my command. Check your PM for a test version before I release it to the less demanding public.

Jeremy

[Uly]

Thanks a lot Jeremy! This version finds a5 38 seconds earlier (or, well, the user is aware of the move 38 seconds earlier).

Stockfish 2.1.1 Gran2c

.13/20	 0:00 	-0.06 	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Bf5 Rf7 6.Ne2 Nc8 7.g6 hxg6 8.Rxg6 Bxd6 9.Rgxd6 Nxd6 10.Rxd6 (405.693) 649
 14/23	 0:01 	-0.02 	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Bf5 Rf7 6.Be6 Rf3 7.Bg4 Re3 8.Bf5 Nc8 9.Bxh7 Nxd6 10.Nf5 Nxf5 11.exf5 (769.608) 703
 15/23	 0:01 	-0.02 	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Bf5 Rf7 6.Be6 Rf3 7.Bg4 Re3 8.Bf5 Nc8 9.Bxh7 Nxd6 10.Nf5 Nxf5 11.exf5 (874.824) 708
 16/23	 0:01 	-0.02 	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Bf5 Rf7 6.Be6 Rf3 7.Bg4 Re3 8.Bf5 Nc8 9.Bxh7 Nxd6 10.Nf5 Nxf5 11.exf5 (1.260.989) 727
 17/23	 0:01 	-0.13++	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Bf5 Rf7 6.Be6 Rf3 7.Bg4 Re3 8.Bf5 Nf8 9.Rgg1 Nc8 (1.452.773) 737
 17/26	 0:02 	-0.08 	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Bf5 Rf7 6.Be6 Rf3 7.Bg4 Rf4 8.c3 Rxg4 9.hxg4 Bxc3 10.Rf2 (1.990.116) 736
 18/26	 0:04 	+0.03--	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Nh5 Nc8 6.Ng7 Nxd6 7.c3 Bc5 8.b4 (3.408.519) 754
 18/26	 0:05 	-0.19++	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Nh5 a5 6.a4 Nc8 7.Bxd7 Kxd7 8.Nf6+ Rxf6 9.gxf6 Nxd6 10.f7 Rf8 (3.812.670) 760
 18/29	 0:06 	 0.00 	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Bf5 Rf7 6.Be6 Rf3 7.Bg4 Re3 8.Bf5 Nf8 9.Rf2 Be1 10.Rg2 Bb4 (4.706.852) 766
 19/29	 0:08 	 0.00 	1...Qf3 2.Qxf3 Rxf3 3.Be2 Rf8 4.Bg4 Bb4 5.Bf5 Rf7 6.Be6 Rf3 7.Bg4 Re3 8.Bf5 Nf8 9.Rf2 Be1 10.Rg2 Bb4 (6.634.040) 773
 20/29	 0:16 	-0.11++	1...a5 2.Nf5 a4 3.Qf3 axb3 4.axb3 Bb4 5.Be2 Ra3 6.Rf1 Bd2 7.Bd3 Bf4+ 8.Kg1 Nc5 (12.852.620) 762
 20/29	 0:21 	-0.22++	1...a5 2.Nf5 a4 3.Qf3 axb3 4.axb3 Bb4 5.Be2 Ra3 6.Rf1 Bd2 7.Bd3 Bf4+ 8.Kg1 Nc5 (16.386.720) 758
 20/29	 0:44 	-0.38++	1...a5 2.Qe3 Bd4 (33.409.696) 753
 20/35	 0:54 	-0.33 	1...a5 2.Nf5 a4 3.Qg4 axb3 4.cxb3 Kc8 5.h4 Kb8 6.h5 Rh8 7.Rc2 Bb4 8.Be2 Qe6 9.Qg1 Ra5 10.Qg3 Rf8 11.Kg2 Nd5 (41.362.295) 755
 21/35	 1:06 	-0.18--	1...a5 2.Nf5 a4 3.h4 axb3 4.axb3 Kc8 5.h5 Rh8 6.Qf3 Kb8 7.Be2 Bb4 8.c3 Bc5 9.b4 (50.015.726) 755
 21/35	 1:22 	-0.14 	1...a5 2.Nf5 a4 3.h4 axb3 4.axb3 Bb4 5.h5 Rh8 6.Qg4 Kc8 7.Rf1 Qe8 8.Nh6 Rf8 9.Nf5 Qd8 10.g6 hxg6 11.Qxg6 Nc5 12.Qg7 Rxf5 13.exf5 (61.922.502) 755
 22/35	 1:24 	-0.14 	1...a5 2.Nf5 a4 3.h4 axb3 4.axb3 Bb4 5.h5 Rh8 6.Qg4 Kc8 7.Rf1 Qe8 8.Nh6 Rf8 9.Nf5 Qd8 10.g6 hxg6 11.Qxg6 Nc5 12.Qg7 Rxf5 13.exf5 (63.376.283) 754
 23/35	 1:35 	-0.14 	1...a5 2.Nf5 a4 3.h4 axb3 4.axb3 Bb4 5.h5 Rh8 6.Qg4 Kc8 7.Rf1 Qe8 8.Nh6 Rf8 9.Nf5 Qd8 10.g6 hxg6 11.Qxg6 Nc5 12.Qg7 Rxf5 13.exf5 (71.691.006) 754

- check your PM.

[User923005]

OK, here's PA_GTB_Gran2b. This should be functionally identical to the 2.0.1 version (and slightly more efficient, in fact). I've left my experiments with Uncombed Coconut's stuff out until I have some way of testing and verifying that they're working. Thanks in advance for any testing, bug reports, etc.

An OSX version is in the archive this time, along with 32- and 64-bit Win builds and the source. As ever, the source is also here:

https://github.com/jeremybernstein/Stockfish_PA_GTB

in single-processor + analysis mode (SF sticks to f7f3 through depth 20). In non-analysis mode, a7a5 becomes interesting at depth 18 through depth 20. If I move f7f3 and go back, f7f3 remains interesting. However, it has the same score (0) as a7a5

By the way, the only thing that UCI_AnalyseMode does is change the king safety weights, so I don't even think that this is terribly relevant. It probably just means that f7f3 is a hair "safer" than a7a5 and/or that the move ordering responds differently to the slightly different king safety evaluations. Anyway, please try to help me see the problem...

Jeremy

Does your code incorporate the Onno Garms changes such as those mentioned in this thread:
http://talkchess.com/forum/viewtopic.ph ... 30&t=38407

[Jeremy Bernstein]

Does your code incorporate the Onno Garms changes such as those mentioned in this thread:
http://talkchess.com/forum/viewtopic.ph ... 30&t=38407

I didn't add it. When did SF2.1.1 come out? *blatter blatter* 10 May. Looks like Marco was convinced on 25 May, so it's not in the SF 2.1.1 codebase, but probably in an upcoming version. If I suddenly have too much free time, I can look into adding this myself. Otherwise, I'll just wait for the next SF release.

Jeremy

[Uly]

before I release it to the less demanding public.

I think it's ready for release

[Jeremy Bernstein]

before I release it to the less demanding public.

I think it's ready for release

Here you are. Stockfish_211_PA_GTB_Gran2c now available for public consumption.

Changes:
- search depth relimited to 100 plies, although pvs might contain up to 256 moves (to accomodate long tablebase pvs) (this was changed to 250 in Stockfish_211_PA_GTB_Gran2a and, while it should have no effect on game play, is nicer as it was)
- fail high / low now output to UCI and logged

Mac, Win32, Win64 & source included. Source maintained at: https://github.com/jeremybernstein/Stockfish_PA_GTB

Enjoy (again - please let me know if you have trouble or questions)
Jeremy

[Uly]

Thanks Jeremy, I have made a sister release thread at Rybka Forum.