Borderline endgame #3

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
User avatar
Nordlandia
Posts: 2561
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Borderline endgame #3

Post by Nordlandia » Mon Nov 25, 2019 2:11 pm

Slight modification over 2r1bk2/pp4pp/1nn5/8/4P3/2N2P2/PP1R2PP/3R2K1 w - - 0 1 | That is presumably winning for white


Long TC game with decent hardware in case on Leela vs Stockfish and vice versa will be appreciated.

User avatar
Nordlandia
Posts: 2561
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: Borderline endgame #3

Post by Nordlandia » Mon Nov 25, 2019 4:21 pm

Not exactly long TC.


peter
Posts: 1853
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Borderline endgame #3

Post by peter » Tue Nov 26, 2019 6:08 am

Hi Jon!
Nordlandia wrote:
Mon Nov 25, 2019 4:21 pm
Not exactly long TC.
Guess it's lost, yet Black could set up some more fight than in your game.
Here with some more hardware- TC:



LC0 was 2xRTX 2080Ti net 49921, SF (Moonfish Jörg Oster) 24 threads of 12x3GHz 8 G hash, 60'+15". 6men Syzygys.
Times and evals stored.
Peter.

User avatar
Nordlandia
Posts: 2561
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: Borderline endgame #3

Post by Nordlandia » Tue Nov 26, 2019 10:50 am

Hi Peter!

I plan to purchase another RTX 2080 Super on Black Friday.

This endgame occured in a 5+5 blitz game between Fat Fritz RTX and SF 10 (default contempt)

What is the analysis. I manually adjudicated it as draw because the eval was stuck on +2 / 2.5. It might be winning with deep analysis.


peter
Posts: 1853
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Borderline endgame #3

Post by peter » Tue Nov 26, 2019 4:56 pm

Nordlandia wrote:
Tue Nov 26, 2019 10:50 am
It might be winning with deep analysis.
Yes, probably.

24 threads of 12x3GHz CPU, 32G hash, 6men- Syzygys, Contempt 0, after some Forward- Backward :

6k1/8/8/2R4p/5B1P/6PK/2p1q1B1/8 w - - 0 1

Analysis by Moonfish 191119 64 POPCNT:

48.Rc3 Qf2 49.Rc8+ Kg7 50.Kh2 Qe2 51.Rc7+ Kg8 52.Bg5 Kf8 53.Rc8+ Kg7 54.Rc6 Kf7 55.Rf6+ Kg7 56.Rf3 Qd1 57.Rf1 Kg6 58.Be3 Qe2 59.Bc1 Kg7 60.Kg1 Qe5 61.Kf2 Qc3 62.Bf3 Qb4 63.Be3 Kg6 64.Rc1 Qa4 65.Bg5 Qd4+ 66.Kg2 Qc4 67.Rg1 Qa2 68.Be4+ Kg7 69.Bc1 Qe6 70.Bf3 Qe8 71.Bg5 Kg8 72.Rc1 Qg6 73.Re1 Kf8 74.Ra1 Kg7 75.Ra7+ Kh8 76.Rc7 Qd3 77.Bf6+ Kg8
+- (6.55) Depth: 53/60 00:16:02 27146MN, tb=1124581

To say it in a Tsvetkovesque way: about 60 moves to mate.
:)
Not because of the output- line, that was just the length of a quick playout with SF (Moonfish) only.

At the end of the output- line above:

6k1/2R5/5B2/7p/7P/3q1BP1/2p3K1/8 w - - 0 1

Analysis by Moonfish 191119 64 POPCNT:

78.Rc8+ Kf7 79.Bg5 Qd1 80.Bf4 Kf6 81.Rc6+ Kg7 82.Bh6+ Kf7 83.Rc5 Ke7 84.Bf4 Qb1 85.Bxh5 Qa2 86.Kh3 Qe6+ 87.Bg4 Qe4 88.Kh2 Qd3 89.Rc7+ Kf8 90.Rc8+ Kf7 91.Bh5+ Kf6 92.Bg5+ Kg7 93.Rc7+ Kf8 94.Kg2 Qd5+ 95.Bf3 Qa2 96.Bc1 Qb1 97.Bh6+ Ke8 98.Bg5 Kf8 99.Rc8+ Kg7 100.Rc6 Kf8 101.h5 Qb2 102.Kh3 Qb5 103.Bf4 Qd3 104.Kg2 c1R 105.Bxc1 Ke7 106.Bg5+ Kd7 107.h6 Qb1 108.Bf4 Ke7 109.Rc7+ Ke6
+- (13.15) Depth: 37/70 00:04:54 7204MN, tb=4174870

And at this one end- postion it's #13, so yet a little longer still, but with probably some unneccessary shuffling.
:)
Peter.

User avatar
Nordlandia
Posts: 2561
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: Borderline endgame #3

Post by Nordlandia » Tue Nov 26, 2019 8:59 pm

Thanks peter. Starting position was

Lc0 is capable of beating SF Dev here as white.

Post Reply