ShashChess

peter · Post by **peter** » Tue Jan 18, 2022 6:30 pm

Jouni wrote: ↑Tue Jan 18, 2022 4:44 pm Peter: yes arasan suite is too easy to detect parameter change. And some positions seems to be incorrect I quess. With HTC and other suites MCTS/Tal is better.

Arasan is ok with single core, I'd say. Vincent Lejeune lets his HTC- suite run that way (single core) yet too normally as you'll probably know, but with 30 minutes/position. I like quick and yet somewhat statistically relevant results with SMP to compare this one hardware- performance too. EloStatTS helps for discrimination of near to each other engines, versions and settings by evaluating the solution- times per position each by each and engine by engine, if you find the time, read the .pdf (pity in German only, as fas as I know) from the link given. Program counts each position solved or unsolved as a single match- result between any two engines competing, time to solution gives more parameters but 1-0, 0-1, 1/2-1/2 only,. So the Elo- rating is quite realistic, of course only for the single very special test. Yet engine- engine- matches nowadays are some kind of played out position- tests too anyhow, results depending on hardware- time (LC0!), opponents and openings more and more too, so what?

Jouni · Post by **Jouni** » Wed Jan 19, 2022 3:36 pm

German is no problem for me! I found an issue with MCTS=1 search:

[fen]4Kb2/5Np1/2p1k1p1/2p5/2PpNnP1/2PP3p/4n3/2rbr1B1 w - - 0 1[/fen]

Analysis by ShashChess 20.2:

1.Nd8+ Ke5 2.Nf7+ Ke6
= (0.00) Depth: 7/7 00:00:00 8kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nd8+
= (0.00) Depth: 8/7 00:00:00 10kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nd8+
= (0.00) Depth: 9/6 00:00:00 12kN
1.Nd8+ Ke5 2.Nf7+ Ke6
= (0.00) Depth: 10/5 00:00:00 13kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nd8+
= (0.00) Depth: 11/6 00:00:00 28kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nd8+
= (0.00) Depth: 12/6 00:00:00 53kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nd8+
= (0.00) Depth: 13/6 00:00:00 91kN
1.Nd8+ Ke5 2.Nf7+ Ke6
= (0.00) Depth: 14/5 00:00:00 103kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nd8+
= (0.00) Depth: 15/6 00:00:00 128kN
1.Nd8+ Ke5 2.Nf7+ Ke6
= (0.00) Depth: 16/5 00:00:00 183kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nfg5+ Ke5
= (0.00) Depth: 17/7 00:00:00 221kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nd8+
= (0.00) Depth: 18/6 00:00:00 264kN
1.Nd8+ Ke5 2.Nf7+ Ke6
= (0.00) Depth: 19/5 00:00:00 320kN
1.Nd8+ Ke5 2.Nf7+ Ke6 3.Nd8+
= (0.00) Depth: 20/6 00:00:00 348kN
1.Nd8+ Ke5 2.Nf7+ Ke6
= (0.00) Depth: 21/5 00:00:00 466kN
1.Nfg5+ Ke5 2.Nf3+
+- (#2) Depth: 22/5 00:00:00 868kN
1.Nfg5+ Ke5 2.Nf7+
+- (#2) Depth: 23/6 00:00:00 924kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 24/6 00:00:00 942kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 25/5 00:00:00 943kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 26/5 00:00:00 943kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 27/5 00:00:00 943kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 28/5 00:00:00 943kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 29/5 00:00:00 944kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 30/6 00:00:00 944kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 31/5 00:00:00 960kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 32/6 00:00:00 987kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 33/5 00:00:00 1018kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 34/6 00:00:00 1052kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 35/6 00:00:00 1120kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 36/6 00:00:00 1202kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 37/6 00:00:00 1279kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 38/5 00:00:00 1360kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 39/6 00:00:00 1428kN
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 40/5 00:00:00 1510kN
1.Neg5+ Kf6 2.Ne4+ Ke6 3.Nfg5+
+- (#3) Depth: 41/5 00:00:00 1600kN
1.Neg5+ Kf6 2.Nh7+
+- (#2) Depth: 42/5 00:00:00 8660kN, tb=66
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 43/30 00:00:02 26270kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 44/6 00:00:02 26270kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 45/5 00:00:02 26271kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 46/6 00:00:02 26271kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 47/5 00:00:02 26271kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 48/6 00:00:02 26271kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 49/5 00:00:02 26271kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 50/6 00:00:02 26272kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 51/5 00:00:02 26272kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 52/6 00:00:02 26272kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 53/5 00:00:02 26272kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 54/6 00:00:02 26272kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 55/6 00:00:02 26273kN, tb=525
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 56/6 00:00:02 26484kN, tb=528
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 57/5 00:00:03 31105kN, tb=828
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 58/5 00:00:03 31145kN, tb=828
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 59/5 00:00:03 31157kN, tb=828
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 60/5 00:00:03 31876kN, tb=835
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 61/6 00:00:03 34160kN, tb=977
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 62/5 00:00:03 34602kN, tb=978
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 63/5 00:00:04 40351kN, tb=2082
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 64/5 00:00:04 42347kN, tb=2246
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 65/6 00:00:04 43835kN, tb=2456
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 66/5 00:00:05 49924kN, tb=3221
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 67/5 00:00:05 51053kN, tb=3256
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 68/5 00:00:05 52023kN, tb=3544
1.Neg5+ Kf6 2.Ne4+
+- (#2) Depth: 69/6 00:00:05 53591kN, tb=4048

5 cores used. Weird

. Position is mate in 15. BTW this happens only sometimes and in Chessbase GUI.

peter · Post by **peter** » Wed Jan 19, 2022 5:42 pm

Jouni wrote: ↑Wed Jan 19, 2022 3:36 pm 5 cores used. Weird . Position is mate in 15. BTW this happens only sometimes and in Chessbase GUI.

Could not reproduce it with 3 trials in Fritz 14. Took 5 threads (of 32 possible) and 1 MCTS thread too. Always #15-output almost at once.
8G hash, Syzygys Probe LImit 0, Persistent Learning Off.

4Kb2/5Np1/2p1k1p1/2p5/2PpNnP1/2PP3p/4n3/2rbr1B1 w - - 0 1

Analysis by ShashChess 20.2:

1.Sd8+ Ke5 2.Sf7+ Ke6
= (0.00) Tiefe: 7/5 00:00:00 11kN
...
1.Seg5+ Kf6 2.Sh7+ Ke6 3.Sxf8+ Kf6 4.Sd7+ Ke6 5.Sxc5+ Kf6 6.Se4+ Ke6 7.Sfg5+ Ke5 8.Sf3+ Ke6 9.Lxd4 Sxd4 10.Sxd4+ Ke5 11.Sxc6+ Ke6 12.Sd8+ Ke5
+- (#15) Tiefe: 17/26 00:00:00 397kN
1.Seg5+ Kf6 2.Sh7+ Ke6 3.Sxf8+ Kf6 4.Sd7+ Ke6 5.Sxc5+ Kf6 6.Se4+ Ke6 7.Sfg5+ Ke5 8.Sf3+ Ke6 9.Lxd4 Sxd4 10.Sxd4+ Ke5 11.Sxc6+ Ke6 12.Sd8+ Ke5 13.Sf7+ Ke6 14.Sfg5+ Ke5
...
1.Seg5+ Kf6 2.Sh7+ Ke6 3.Sxf8+ Kf6 4.Sd7+ Ke6 5.Sxc5+ Kf6 6.Se4+ Ke6 7.Sfg5+ Ke5 8.Sf3+ Ke6 9.Lxd4 Sxd4 10.Sxd4+ Ke5 11.Sxc6+ Ke6 12.Sd8+ Ke5 13.Sf7+ Ke6 14.Sfg5+ Ke5 15.d4#
+- (#15) Tiefe: 52/30 00:00:10 122MN

I use bmi2 compile (a little faster at bench at my Ryzen). Did you maybe download before 16.1.20h? It's said, there had been issues with avx2- compile of 20.2 before renewing of download- link next day.

Dicaste · Post by **Dicaste** » Wed Jan 19, 2022 11:08 pm

Sometimes it throws exception in Fritz(Chessbase) gui. Is it happening only for me ?

peter · Post by **peter** » Thu Jan 20, 2022 12:46 am

Dicaste wrote: ↑Wed Jan 19, 2022 11:08 pm Sometimes it throws exception in Fritz(Chessbase) gui. Is it happening only for me ?

peter wrote: ↑Wed Jan 19, 2022 5:42 pm Did you maybe download before 16.1.20h? It's said, there had been issues with avx2- compile of 20.2 before renewing of download- link next day.

Paloma · Post by **Paloma** » Thu Jan 20, 2022 1:35 am

Hi Peter,

from your +20 positions above:
8/1pp2k2/p2p3p/3n1KP1/5P1P/8/1P1B4/8 w - - bm Ke4; id "Nebenvariante 41.Ke4!, 9";

what is the source of this pos?

peter · Post by **peter** » Thu Jan 20, 2022 1:53 am

Paloma wrote: ↑Thu Jan 20, 2022 1:35 am Hi Peter,

from your +20 positions above:
8/1pp2k2/p2p3p/3n1KP1/5P1P/8/1P1B4/8 w - - bm Ke4; id "Nebenvariante 41.Ke4!, 9";

what is the source of this pos?

According to my database where I picked it out, that's a side- line to Lasker E.- Lipke P., Breslau 1889.
Sorry that some of the positions have incomplete sources in .epd, .pgn to .epd- tool did that here and there and I was too lazy to edit.

Jouni · Post by **Jouni** » Thu Jan 20, 2022 5:36 pm

ShashChess MCTS1 + Tal is really impressive solver. In Arasan suite it solves 199/200 in first or second run fast! It only misses position 189 after 10 minutes. But default mode solves.

Dicaste · Post by **Dicaste** » Fri Jan 21, 2022 12:48 am

I downloaded new bmi version again. Yet it still throws an exception.

My settings are something like this. (4096 MB) (MCTS doesn't affect the result. It still throws no matter what.

It occurs in some specific positions. I couldn't figured out yet but it can't finish marathon test suite or even tournament matches.

[fen]r1b1rnk1/1p3pb1/1qpp2p1/2n5/2PNP2p/1PN3PP/3R1PBK/BR1Q4 b - - 0 1[/fen]

Vinvin · Post by **Vinvin** » Fri Jan 21, 2022 12:54 am

peter wrote: ↑Tue Jan 18, 2022 12:20 am ... So 222 positions all together, "new" as for (to me) known suites are only these 20:

Code: Select all

6k1/p1p1p1p1/P5Pq/7P/8/5R2/1P1P4/5K1B w - - bm Rf5; id "Stadlhofer Anton, 1";
3q3N/1p5k/5B2/8/1p2p1Rp/1P5K/p3PP2/8 w - - bm Rg7+; id "M. Garcia - P. Krug , Studie 2015, 2";
n2Bqk2/5p1p/Q4KP1/p7/8/8/8/8 w - - bm Qc8; id "Matous, Mario, 3";
k7/7p/2p5/2rp1PP1/pp6/7p/5K1P/4R3 w - - bm g6; id "Zepler, E.E., 4";
7k/pp1pBp1N/1p3Pp1/b3K1PR/P3P2p/R1n4P/2P1N1P1/3q4 w - - bm Rxc3; id "Josten, Gerhard, 5";
k7/P4pp1/1P6/2p1P3/2P1n3/p5pB/P4n2/K5b1 w - - bm e6; am Bg2; id "Hörning, Gerd, 6";
5N1n/3P1p1p/5P1k/1Pp1P2p/4KP1P/2p5/p4b2/5N2 w - - bm Ne3; id "Formela, Michal , 7";
r1b2rk1/3n1pb1/2p1p1p1/p5P1/Pp2B1NQ/1q4P1/2R2P2/3R2K1 w - - bm Rxd7; id "TCEC 2019, 8";
8/1pp2k2/p2p3p/3n1KP1/5P1P/8/1P1B4/8 w - - bm Ke4; id "Nebenvariante 41.Ke4!, 9";
1r1q1r1k/5pp1/3p1b2/pp1BpPB1/1n2P1QP/1PN5/1PPK4/R7 w - - bm Ne2; id "corr LIPEAD40 (PER) , 10";
r3nrk1/1bqp1ppp/p1n1p3/1p4PQ/3NP3/2N2R2/PPP1B2P/3RK3 w - - bm Nd5; id "Anton Stadlhofer, 11";
r1b2rk1/pp3p2/2p3pp/2n5/8/q3RN2/P4PPP/2RQ1BK1 b - - bm Qa4; id "Chessbase online, 12";
n6r/b7/1R4R1/2k5/2pppppp/8/1PPPPPP1/r4NK1 w - - bm b4+ Rgc6+; id "Korolkov W.A., 13";
r4r2/b2q2pk/3p2np/N2pp3/PP1P4/2P3PP/1B3P2/R3RQK1 b - - bm Rf5; id "Chessbase online , 14";
2r2krb/1qp2p2/2NPpN2/1P5B/p2Bp2p/1p5b/7p/R3K2Q w Q - bm Nh7+; id "Krug P., Garcia M.., 15";
8/7P/1P5B/2B1Q1n1/3nn2P/1PRnk1nR/3nnnK1/2B1nQBn w - - bm Qxe1; id "Goldsteen H., Horse Connection, 16";
r2n1r1k/1pp5/3ppn2/P2Pp2p/2P1P1pq/1Q2R2P/R2N1PP1/4N1K1 w - - bm g3; id "WC30/final, 17";
2b5/1pr4p/3bp1pk/1p6/1PpN2PP/K1P1n3/P3N1R1/3R4 w - - bm Nxb5; id "Mihai Neghina, 18";
3r2k1/p4pP1/1ppr4/5Bp1/P2qPnQP/4R3/5P2/5RK1 w - - bm e5; id "Tauber-Sikorsky ICCF 2018, 19";
4q1kr/p6p/1prQPppB/4n3/4P3/2P5/PP2B2P/R5K1 w - - bm Qxe5; id "Gusev-Auerbach, Molniya 1946, 20";

Hi Peter,
about half of these positions are well known for some years. And they are too easy for current engines.

Vincent

ShashChess

Re: ShashChess

Re: ShashChess

Re: ShashChess

Re: ShashChess

Re: ShashChess

Re: ShashChess

Re: ShashChess

Re: ShashChess

Re: ShashChess

Re: ShashChess