Sting-sf-5 release

[lech]

As earlier, my project Sting is based on Stockfish 2.1.1 free source code.
I found and corrected two seriouse bugs in the code of Stockfish, and they were very important for all my work.
For this, I removed links to all the previous versions of Sting.
Practically the version of Sting sf 5 is the same as previous 4.8.4 but "only" with these corrections.
The strenght of both should be similar and, in my opinion, it has no sense to test it.

A few words about my project.
I stated this to prove that engines can solve very difficult chess positions (the horizon effect) and to save a high ELO.
For Sting, an ELO question (gain) aplies only to the difference between Sting and the parent Stockfish 2.1.1 .
I solved many code questions and the problem with horizon effect (appears mainly in endgames), is already no motivation for scientists to create any artifical intelligence.
I see some new questions for my project but now it is not too easy to find chess positions out of range for engines.

Link: http://www.mediafire.com/download/utuh6 ... g-sf-5.zip
It contains: win32, x64 and x64(Win XP) compilations.

[Frank Quisinsky]

Good News!
Give me an order if I have to test a new version of Sting SF. Must be an official version for me (I am not test private engines or alpha, beta, cesar test versions).

For the moment I hold version 4.8 in my rating list.

THANKS!!
I like your work!

Best
Frank

[Paloma]

Thank's Marek for this Version of Sting!

Would you please tell us what kind of bug you found?
And what effects in analysis they have?

[lech]

Thank's Marek for this Version of Sting!

Would you please tell us what kind of bug you found?
And what effects in analysis they have?

The first bug leads to crashing of Sting on a very high depths (too smal memory for ss).
The second doesn't allow to find the correct move (futilityValue goes over beta after a possible weak capture).
It is described with examples here: http://talkchess.com/forum/viewtopic.php?t=55411

BTW. Sting sf 5 is not a beta version.

[fishpov]

Hi Marek this is good news !
But you say that you are near to finish, I really hope you will continue because you are going to realize what I considered impossible before.
I think you have at least 5 years of advances on other engines for these kind of fortress or impossible position.
Thanks a lot for your very good job !!

[lech]

Hi Marek this is good news !
But you say that you are near to finish, I really hope you will continue because you are going to realize what I considered impossible before.
I think you have at least 5 years of advances on other engines for these kind of fortress or impossible position.
Thanks a lot for your very good job !!

Indeed I am very close to end my work. Unfortunately, it happens.

Some information about the last version 4.8.4:

Sting SF 4.8.4 was downloaded 1139 times (previuos 4.7 - 997).

CCRL 40/40 (complete list).
Sting SF 4.8.4 64-bit 4CPU 3147 285 games
Stockfish 2.1.1 64-bit 4CPU 3128 406 games
Sting SF 4.8.4 64-bit 3076 370 games
Stockfish 2.1.1 64-bit 3069 1186 games

Many thanks for CCRL team!

At the end my liked funny position for engines. I set it myself a few years ago.
[d]rk6/p1p5/Bp2p3/1P6/q2B1P2/4K3/8/8 b - - to win it needs to sacrifice Queen (now or later) for one of white Bishops ONLY (to make Rook a8 free and to realize passed pawns).

[Lanzo]

Marek, I hope you like challenges because this one, will probably be the toughest you'll ever face.

The Arasan Testsuite is arguably the strongest tactical testsuite of all, and is very essential in determining an engine's calculation reflexes when exposed to such positions. You can find it here; http://www.arasanchess.org/arasan18.epd

Currently, Houdini is the king of tactical positions. Stockfish solves only about 134/250 of the problems given one minute to think in the Arasan testsuite, while Houdini solves 203/250 of the problems! It is practically impossible for an engine to solve all the problems in the test suite and the day that happens, will be a great day. The challenge is, can YOU make it happen?

[Paloma]

But on this Pos. here Sting SF 5 still sticks on Depth 29 (more than 4 hours, reaches Depth 29 in 15 min.)

Especially the pv and the kN/s changed never since then (after 15 min 1469 kN/n after 4 h still 1469)



Please Click on the Picture to enlarge.

I used 1 core and Taskmanager shows 50% utilization, but no progess with Sting's output.
It seems Sting hangs in a *infinity loop*

[zullil]

Stockfish solves only about 134/250 of the problems given one minute to think in the Arasan testsuite,

This is completely incorrect, unless you have crippled Stockfish or are running it on an abacus.

[BBauer]

This position seams not to be very difficult.
My slightly modified stockfish gives the right move just from the start.
Kind regards
Bernhard

Code: Select all

[D]rk6/p1p5/Bp2p3/1P6/q2B1P2/4K3/8/8 b - - 0 1[D]

St-mod-15-04-03:
 8/14	00:00	 10k	532k	+10,04	Da4-b3+ Ke3-e4 Db3-b4 Ke4-e3
 9/16	00:00	 11k	564k	+10,04	Da4-b3+ Ke3-e4 Db3-b4 Ke4-e3
 10/19	00:00	 18k	660k	+10,04	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 11/21	00:00	 25k	743k	+10,34	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 12/24	00:00	 38k	879k	+10,38	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 13/27	00:00	 59k	1.090k	+10,44	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 14/27	00:00	 83k	1.239k	+10,43	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 15/27	00:00	 93k	1.256k	+10,43	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 16/28	00:00	 121k	1.249k	+10,43	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 17/28	00:00	 174k	1.362k	+10,39	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 18/28	00:00	 213k	1.409k	+10,44	Da4-b3+ Ke3-e4 Db3-b1+ Ke4-e3
 19/36	00:00	 492k	1.601k	+10,77	Da4-b3+ Ke3-e2 Db3-c2+ Ke2-e3
 20/36	00:00	 627k	1.722k	+10,90	Da4-b3+ Ke3-e2 Db3-c2+ Ke2-e3
 21/37	00:00	 713k	1.748k	+10,96	Da4-b3+ Ke3-e2 Db3-c2+ Ke2-e3
 22/37	00:00	 815k	1.567k	+11,03	Da4-b3+ Ke3-e2 Db3-c2+ Ke2-e3
 23/37	00:00	 1.008k	1.363k	+11,03	Da4-b3+ Ke3-e2 Db3-c2+ Ke2-e3
 24/39	00:01	 1.518k	1.202k	+11,15	Da4-b3+ Ke3-e2 Db3-c2+ Ke2-e3
 25/39	00:01	 1.814k	1.112k	+11,15	Da4-b3+ Ke3-e2 Db3-c2+ Ke2-e3
 26/44-	00:03	 4.040k	1.341k	+10,58	Da4-b3+ Ke3-e2
 26/44-	00:03	 4.306k	1.365k	+10,27	Da4-b3+ Ke3-e2
 26/52+	00:03	 4.545k	1.412k	+10,51	Da4-b3+
 26/52+	00:03	 4.656k	1.429k	+11,21	Da4-b3+
 26/52	00:03	 5.336k	1.532k	+11,21	Da4-b3+ Ke3-f2 Db3-c2+ Kf2-e3
 27/52+	00:03	 5.526k	1.531k	+11,27	Da4-b3+
 27/52	00:03	 6.032k	1.567k	+11,27	Da4-b3+ Ke3-f2 Db3-c2+ Kf2-e3
 28/52	00:04	 7.441k	1.707k	+11,27	Da4-b3+ Ke3-f2 Db3-c2+ Kf2-e3
 29/52	00:05	 9.493k	1.798k	+11,27	Da4-b3+ Ke3-f2 Db3-c2+ Kf2-e3
 30/52-	00:08	 14.563k	1.697k	+11,21	Da4-b3+ Ke3-f2
 30/52-	00:11	 19.455k	1.741k	+11,15	Da4-b3+ Ke3-f2
 30/52-	00:16	 31.092k	1.834k	+9,93	Da4-b3+ Ke3-f2
 30/64	00:18	 33.942k	1.880k	+10,05	Da4-b3+ Ke3-f2 Db3-c2+
 31/64	00:19	 38.415k	1.924k	+10,05	Da4-b3+ Ke3-f2 Db3-c2+
 32/64	00:22	 45.229k	2.012k	+10,05	Da4-b3+ Ke3-f2 Db3-c2+
 33/64+	00:23	 48.203k	2.036k	+10,11	Da4-b3+
 33/64+	00:23	 48.787k	2.045k	+10,17	Da4-b3+
 33/64+	00:48	 126.689k	2.627k	+45,82	Da4-b3+
 33/64	00:55	 149.846k	2.710k	+51,99	Da4-b3+ Ke3-f2 Db3-c2+
 34/64-	00:58	 158.812k	2.696k	+51,92	Da4-b3+ Ke3-d2
 34/64+	00:59	 159.621k	2.698k	+51,99	Da4-b3+
 34/64+	01:06	 186.295k	2.800k	+56,54	Da4-b3+
 34/64-	01:08	 192.207k	2.811k	+55,25	Da4-b3+ Ke3-e2
 34/64-	01:11	 201.077k	2.824k	+46,48	Da4-b3+ Ke3-e2
 34/64+	01:11	 201.136k	2.824k	+50,49	Da4-b3+
 34/64	01:13	 208.195k	2.840k	+52,03	Da4-b3+ Ke3-e2 Db3-c2+
 35/64+	01:13	 208.616k	2.839k	+52,09	Da4-b3+
 35/64+	01:13	 209.223k	2.838k	+52,15	Da4-b3+
 35/64+	01:20	 232.701k	2.874k	+105,70	Da4-b3+
 35/64-	01:31	 265.416k	2.909k	+92,14	Da4-b3+ Ke3-e2
 35/64	01:33	 271.487k	2.911k	+56,80	Da4-b3+ Ke3-e2 Db3-c2+
 36/64+	01:33	 272.287k	2.910k	+56,86	Da4-b3+
 36/64+	01:33	 272.574k	2.911k	+56,92	Da4-b3+
 36/64+	01:33	 273.077k	2.911k	+57,36	Da4-b3+
 36/64-	01:35	 278.876k	2.915k	+57,15	Da4-b3+ Ke3-e2
 36/64+	01:35	 279.078k	2.915k	+57,44	Da4-b3+
 36/64-	01:39	 291.347k	2.915k	+57,08	Da4-b3+ Ke3-e2
 36/64+	01:40	 291.732k	2.916k	+57,61	Da4-b3+
 36/64+	01:41	 297.586k	2.919k	+108,18	Da4-b3+
 36/64	01:59	 345.261k	2.893k	+123,34	Da4-b3+ Ke3-e2 Db3-c2+ Ke2-e3 Dc2-c1+ Ke3-e4 Dc1-e1+ Ke4-d3 c7-c5 Ld4-e5+ De1xe5 f4xe5 Kb8-c7 Kd3-c3 Ta8-f8 Kc3-d3 Tf8-f3+ Kd3-e2 Tf3-g3 Ke2-d1 c5-c4 Kd1-d2 c4-c3+ Kd2-c2 Tg3-f3 Kc2-b1 c3-c2+ Kb1-b2 Tf3-a3 Kb2xa3 c2-c1D+ Ka3-a2 Dc1-c2+ Ka2-a3 Dc2-c3+ Ka3-a2 Dc3-a5+ Ka2-b2 Da5xa6 Kb2-b1 Da6xb5+ Kb1-c2 Db5xe5 Kc2-b3 De5-c3+ Kb3-a2 Dc3-b2+ Ka2xb2