Komodo Dragon vs. GM Perelshteyn, knight odds match.

[lkaufman]

Revisiting my thoughts on this match, my opinion has now changed given that the match does not start until Nov 6th and that the GM does have a version of Komodo to practice with, I now think he should have good chances to win the match. I don't believe I gave enough thought to this little detail.

Since he is practicing with Komodo Dragon 2.5.1, he only need ChrisW opening book and Mr. Kaufman 32 core Threadripper, but I strongly recommend Mr. Kaufman to set his Komodo Dragon xxx.x to play at (5'+2") instead of (3' + 2")

I've run enough simulations to say that using more than 16 threads or more than 3' +2" (with 16 threads) is totally useless for knight odds. Imagine if we had a 32 man tablebase (from a million years in the future!). How would that help us in a knight odds match? Sure, if we got down to less than a pawn deficit in a game it might see a weird draw that normal search can't find, but against a knight-odds human opponent there is no point in assuming that he will play perfectly, it's better not to see every little detail in the far future that he will never even think of. If we get within drawing range we will surely not lose even if we play imperfectly, and if we reach a position where a win is possible we will probably win even if we miss the first win. Knight odds chess requires a different mental model than normal chess. Also, the ChrisW opening book is of no use for a human preparing, it is only for getting variety for computer vs computer matches. The positions (mostly) won't actually occur in real knight odds games.

If that is the case how come you will use a 32 core Threadripper instead of a 16 threads >

If we go with 16 threads then indeed I could use my fast 16 core machine instead. But everything is set up on my threadripper now, so there is no advantage to switching, and I might decide to use maybe 24 threads, depending on testing.

By now you probably have done several testing with 16 threads and with 24 threads; which seems to be more effective at Knight Odds against an engine of similar strength Gaviota-win32-0.84 = GM Perelshteyn ?

So far it seems 16 threads is best for this, but there are many ways to limit the search, by time, by depth, by Skill level, or by threads. So it's not a simple question, I'm trying to find the best combination. I'm hoping that we can draw the match, but I'm not predicting that.

Since he is practicing with Komodo Dragon 2.5.1 at 15'+10" , there is a way to get a least a draw, a day before the match offer him to play at 10'+10" instead

I was fooling around, but since GM Perelshteyn is practicing with Komodo Dragon 2.5.1 the best that Komodo Dragon xxx.x can do is to get a Draw. But even if Tinsley was allowed to practice against Chinook, a checkers-playing program programmed by Jonathan Schaeffer, he was NOT going to beat Chinnook, the best that he could have done was draw several games

Note: I do NOT believe that Chinook won the match since Chinook became the first computer program in history to win a human world championship by drawing 6 games, but only because his stomach hurt. The pain was keeping him up all night. After six games—all draws—he needed to see a doctor But Schaeffer was crushed. He’d devoted years of his life to creating a program that could beat the best checkers player ever, and just as he was about to realize this dream, Tinsley quit. Seven months later, Tinsley died, never having truly lost a match to Chinook.
https://www.theatlantic.com/technology/ ... rs/534111/

Note that a drawn match in this case does not imply six draws. Knight odds is not all that drawish, most typically either White stays a piece down and loses, or else wins it back (or the equivalent) somehow and goes on to win. Of course there will be perpetual checks and drawn endgames too. If the match is indeed drawn, I think the most likely scenario is two wins each plus two draws.

[Chessqueen]

[fen]bbqrknnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQk - 0 1 [/fen]

FEN: bbqrknnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQk - 0 1

Dragon-2.5-64bit-avx2:
1 00:00 2 2k -0.56 Ng1-f3
2 00:00 34 34k -0.48 Ng1-f3 c7-c5
3 00:00 111 111k -0.80 Ng1-f3 d7-d5 e2-e3
4 00:00 250 250k -0.79 Ng1-f3 Nf8-g6 h2-h4 c7-c5
5- 00:00 302 302k -0.94 Ng1-f3 Nf8-g6
5 00:00 525 525k -1.02 Ng1-f3 d7-d5 e2-e3 Ng8-f6 c2-c4
6 00:00 785 785k -1.05 Ng1-f3 c7-c5 c2-c4 Ng8-f6 e2-e3 b7-b6
7 00:00 1k 980k -1.05 Ng1-f3 c7-c5 c2-c4 Ng8-f6 e2-e3 b7-b6 b2-b3
8 00:00 2k 1,125k -1.05 Ng1-f3 c7-c5 c2-c4 Ng8-f6 g2-g3 b7-b6 Bf1-g2 d7-d5 c4xd5
9 00:00 3k 1,184k -1.05 Ng1-f3 c7-c5 c2-c4 Ng8-f6 g2-g3 b7-b6 Bf1-g2 d7-d5 c4xd5 Ba8xd5 Nb1-c3
10 00:00 4k 1,273k -1.05 Ng1-f3 c7-c5 c2-c4 Ng8-f6 g2-g3 b7-b6 Bf1-g2 d7-d5 c4xd5 Ba8xd5 Nb1-c3
11 00:00 5k 1,317k -1.05 Ng1-f3 c7-c5 c2-c4 Ng8-f6 g2-g3 b7-b6 Bf1-g2 d7-d5 c4xd5 Ba8xd5 Nb1-c3 Bd5-c6 O-O
12 00:00 8k 1,341k -1.05 Ng1-f3 c7-c5 c2-c4 Ng8-f6 Nb1-c3 b7-b6 a2-a4 Nf8-g6 a4-a5 b6-b5 c4xb5 d7-d5 d2-d4 c5xd4
13 00:00 13k 1,260k -1.05 Ng1-f3 c7-c5 c2-c4 Ng8-f6 Nb1-c3 b7-b6 a2-a4 Nf8-g6 a4-a5 b6-b5 c4xb5 d7-d5 b5-b6 a7xb6 a5xb6 Ba8-c6 Nc3-a4 d5-d4 Na4xc5
14 00:00 18k 1,302k -1.06 Ng1-f3 c7-c5 c2-c4 Ng8-f6 Nb1-c3 b7-b6 a2-a4 Nf8-g6 a4-a5 b6-b5 c4xb5 d7-d5 b5-b6 a7xb6 a5xb6
15- 00:00 38k 1,233k -1.13 Ng1-f3 c7-c5
15 00:00 88k 1,150k -1.09 d2-d4 d7-d5 Bc1-f4 Nf8-g6 Bf4-g3 h7-h5 h2-h3 c7-c5 Bg3xb8 Qc8xb8 e2-e3 c5xd4 Qd1xd4 e7-e5 Qd4-a4+ Ke8-f8 Ng1-f3 Ng8-f6 Qa4-a3+ Kf8-g8
16- 00:00 134k 1,025k -1.16 d2-d4 d7-d5
16 00:00 164k 970k -1.14 Ng1-f3 c7-c5 c2-c4 Ng8-f6 Nb1-c3 b7-b6 a2-a4 Nf8-g6 a4-a5 b6-b5 c4xb5 d7-d5 d2-d3 d5-d4 Nc3-a4 e7-e5 e2-e4 d4xe3/ep Bc1xe3
16 00:00 203k 999k -1.11 b2-b4 e7-e5 Ng1-f3 Nf8-g6 Bc1-b2 c7-c6 e2-e3 d7-d5 c2-c4 Ng8-f6 c4xd5 Nf6xd5 h2-h4 Nd5xe3 f2xe3 e5-e4 Nb1-c3 e4xf3
17- 00:00 230k 984k -1.18 b2-b4 e7-e5
17 00:00 309k 1,004k -1.14 Ng1-f3 c7-c5 c2-c4 Ng8-f6 Nb1-c3 b7-b6 g2-g3 d7-d5 c4xd5 Nf6xd5 Nc3xd5 Ba8xd5 Bf1-g2 Nf8-g6 O-O h7-h5 h2-h4 O-O d2-d3 Ng6-e5
18- 00:00 468k 961k -1.21 Ng1-f3 c7-c5
18+ 00:00 570k 943k -1.07 b2-b4 e7-e5 Bc1-b2
18+ 00:00 622k 942k -0.98 b2-b4 e7-e5 Bc1-b2
18 00:00 717k 937k -0.99 b2-b4 Ng8-f6 Ng1-f3 b7-b6 c2-c4 Nf8-g6 Bc1-b2 c7-c5 b4xc5 b6xc5 e2-e3 e7-e5 d2-d3 O-O Bf1-e2 Rf8-e8 Nb1-c3 d7-d5 c4xd5 Nf6xd5 O-O
19- 00:00 750k 934k -1.06 b2-b4 Ng8-f6
19- 00:00 830k 935k -1.15 b2-b4 Ng8-f6
19 00:00 850k 940k -1.14 c2-c4 Ng8-f6 Nb1-c3 b7-b6 Ng1-f3 c7-c5 g2-g3 d7-d5 c4xd5 Nf6xd5 Bf1-g2 Nf8-g6 O-O Nd5-b4 d2-d3 O-O a2-a3 Nb4-c6 Bc1-e3 e7-e5 Be3-g5 f7-f6 Bg5-d2
20+ 00:00 904k 942k -1.07 c2-c4 Ng8-f6 Nb1-c3
20- 00:00 927k 948k -1.14 c2-c4 Ng8-f6
20+ 00:01 1,047k 950k -1.11 c2-c4 Ng8-f6 Nb1-c3
20 00:01 1,057k 955k -1.13 c2-c4 Ng8-f6 Nb1-c3 b7-b6 Ng1-f3 c7-c5 g2-g3 d7-d5 c4xd5 Nf6xd5 Nc3xd5 Ba8xd5 Bf1-g2 h7-h5 b2-b4 c5xb4 Bc1-b2 Nf8-e6 Qd1-a4+ Qc8-d7
21 00:01 1,159k 958k -1.09 c2-c4 Ng8-f6 Nb1-c3 b7-b6 Ng1-f3 c7-c5 g2-g3 d7-d5 c4xd5 Nf6xd5 Nc3xd5 Ba8xd5 Bf1-g2 Nf8-g6 O-O h7-h5 d2-d4 Bd5xf3 e2xf3 Rd8xd4 Qd1-e2 h5-h4 Bc1-e3 h4xg3 h2xg3
21+ 00:01 1,527k 921k -1.06 b2-b4 Ng8-f6 Ng1-f3
21+ 00:01 1,629k 925k -0.97 b2-b4 Ng8-f6 Ng1-f3
21 00:01 1,681k 921k -1.00 b2-b4 Ng8-f6 Ng1-f3 b7-b6 c2-c4 Nf8-g6 Bc1-b2 c7-c5 b4xc5 b6xc5 e2-e3 O-O Bb2xf6 g7xf6 Nb1-c3 e7-e6 Bf1-e2 f6-f5 Ra1-b1 Kg8-h8 Qd1-c2 Ng6-e5 Nc3-b5 f7-f6 Nf3xe5 f6xe5
22+ 00:04 3,501k 857k -0.93 b2-b4 Ng8-f6 Ng1-f3
22- 00:04 3,881k 852k -1.00 b2-b4 Ng8-f6
22+ 00:04 4,144k 848k -0.97 b2-b4 Ng8-f6 Ng1-f3
22- 00:05 4,330k 849k -1.04 b2-b4 Ng8-f6
22+ 00:05 4,468k 845k -0.99 b2-b4 Ng8-f6 Ng1-f3
22 00:05 4,600k 844k -1.03 b2-b4 Ng8-f6 Ng1-f3 b7-b6 c2-c4 d7-d5 c4xd5 Ba8xd5 Nb1-c3 Bd5-b7 g2-g3 Nf8-e6 Bf1-g2 Ne6-d4 O-O h7-h5 e2-e4 Nd4xf3+ Bg2xf3 c7-c5 b4xc5 Qc8xc5 Bc1-b2 Qc5-b4 Qd1-b3 Qb4xb3 a2xb3 Rd8xd2
23+ 00:07 6,028k 830k -0.96 b2-b4 Ng8-f6 Bc1-b2
23- 00:07 6,404k 829k -1.03 b2-b4 Ng8-f6
23+ 00:08 6,821k 826k -1.00 b2-b4 Ng8-f6 Bc1-b2
23- 00:08 7,070k 827k -1.07 b2-b4 Ng8-f6
23 00:08 7,387k 829k -1.06 b2-b4 Ng8-f6 Bc1-b2 b7-b6 c2-c4 Nf8-g6 h2-h4 h7-h5 e2-e3 d7-d5 c4xd5 Ba8xd5 Qd1-a4+ c7-c6 Nb1-c3 Bd5-e6 Ra1-c1 Qc8-d7 Ng1-f3 O-O d2-d4 Be6-f5 b4-b5 c6-c5 Bf1-e2 c5xd4 Nf3xd4 Rf8-e8
24+ 00:10 8,446k 828k -0.99 b2-b4 Ng8-f6 Bc1-b2
24 00:10 8,812k 829k -0.98 b2-b4 Ng8-f6 Bc1-b2 b7-b6 c2-c4 Nf8-g6 h2-h4 h7-h5 Ng1-f3 c7-c5 b4xc5 b6xc5 e2-e3 Qc8-c6 Nb1-c3 O-O Qd1-c2 e7-e6 Nf3-g5 d7-d5 Bf1-e2 a7-a6 f2-f4 d5-d4 Be2-f3
25+ 00:13 11,124k 827k -0.91 b2-b4 Ng8-f6 Bc1-b2
25- 00:14 11,765k 825k -0.98 b2-b4 Ng8-f6
25+ 00:14 12,369k 825k -0.95 b2-b4 Ng8-f6 Bc1-b2
25- 00:15 12,731k 824k -1.02 b2-b4 Ng8-f6
25+ 00:17 14,491k 818k -0.97 b2-b4 Ng8-f6 Ng1-f3
25- 00:18 15,135k 818k -1.10 b2-b4 Ng8-f6
25+ 00:20 16,705k 817k -1.00 b2-b4 Ng8-f6 Ng1-f3
25 00:20 17,026k 820k -0.96 b2-b4 Ng8-f6 Ng1-f3 Nf8-g6 Bc1-b2 O-O Bb2xf6 g7xf6 c2-c4 b7-b6 Nb1-c3 c7-c5 b4xc5 b6xc5 g2-g3 Bb8-e5 Ra1-c1 Be5xc3 Rc1xc3 Ba8xf3 e2xf3 d7-d5 Bf1-e2 d5xc4 O-O Rd8-d4 f3-f4 Rf8-d8 Be2xc4 Rd4xd2
26- 00:24 19,926k 817k -1.03 b2-b4 Ng8-f6
26+ 00:28 23,263k 814k -1.01 b2-b4 Ng8-f6 Ng1-f3
26- 00:29 23,747k 816k -1.06 b2-b4 Ng8-f6
26+ 00:29 24,343k 817k -1.02 b2-b4 Ng8-f6 Ng1-f3
26 00:29 24,464k 817k -1.05 b2-b4 Ng8-f6 Ng1-f3 Nf8-g6 Bc1-b2 O-O Bb2xf6 e7xf6 g2-g3 b7-b6 Bf1-g2 c7-c5 O-O d7-d5 c2-c3 h7-h5 d2-d4 h5-h4 Nb1-d2 h4-h3 Bg2-h1 Bb8-d6 a2-a3 Rf8-e8 e2-e3 c5-c4 a3-a4
27- 00:35 28,754k 816k -1.12 b2-b4 Ng8-f6
27+ 00:46 37,513k 811k -1.10 b2-b4 Ng8-f6 Ng1-f3
27 00:46 37,612k 811k -1.10 b2-b4 Ng8-f6 Ng1-f3 Nf8-g6 Bc1-b2 O-O Bb2xf6 e7xf6 g2-g3 b7-b6 Bf1-g2 c7-c5 b4xc5 b6xc5 O-O d7-d5 d2-d4 h7-h5 h2-h4 Rf8-e8 e2-e3 Ng6-e7 d4xc5 Bb8xg3 f2xg3 Ne7-f5 Nf3-d4 Nf5xe3 Qd1-d2 Ne3xf1
28+ 00:49 39,937k 810k -1.03 b2-b4 Ng8-f6 Ng1-f3
28 00:50 40,724k 810k -1.05 b2-b4 Ng8-f6 Ng1-f3 Nf8-g6 Bc1-b2 O-O Bb2xf6 e7xf6 g2-g3 b7-b6 Bf1-g2 c7-c5 b4xc5 b6xc5 O-O d7-d5 d2-d4 h7-h5 h2-h4 Rf8-e8 e2-e3 Ng6-e7 d4xc5 Bb8xg3 f2xg3 Ne7-f5 Nf3-d4 Nf5xe3 Qd1-d2 Ne3xf1 Kg1xf1 Qc8xc5 Nb1-c3 g7-g6 Ra1-b1

Stockfish_14_x64_bmi2:
NNUE evaluation using nn-3475407dc199.nnue enabled
1/1 00:00 20 20k -0.77 d2-d4
2/2 00:00 45 45k -0.99 d2-d4 c7-c6
3/3 00:00 95 95k -0.72 d2-d4 c7-c6 e2-e3
4/4 00:00 647 324k -1.45 b2-b4 Ng8-f6 e2-e3 c7-c6
5/5 00:00 1k 500k -1.43 b2-b3 e7-e5 e2-e3 Ng8-f6 Ng1-f3
6/6 00:00 1k 528k -1.27 b2-b3 e7-e5 Bc1-b2 c7-c5 e2-e3
7/7 00:00 2k 790k -1.25 b2-b4 Ng8-f6 Bc1-b2 d7-d6 Bb2xf6 g7xf6 e2-e3
8/10 00:00 6k 921k -1.35 b2-b4 Ng8-f6 Bc1-b2 b7-b6 e2-e3 c7-c5 b4xc5 b6xc5 Ng1-f3
9/13 00:00 10k 1,002k -1.27 d2-d4 Ng8-f6 e2-e3 Nf8-g6 Ng1-f3 c7-c5 Bf1-d3 c5xd4
10/12 00:00 30k 1,058k -1.37 b2-b4 e7-e5 Bc1-b2 c7-c6 d2-d4 e5-e4 e2-e3 Ng8-f6 c2-c4
11/15 00:00 59k 1,075k -1.42 b2-b4 b7-b6 Bc1-b2 Ng8-f6 c2-c4 c7-c5 e2-e3 c5xb4 a2-a3 b4xa3 Bb2xf6 g7xf6
12/16 00:00 86k 1,077k -1.43 b2-b4 b7-b6 Bc1-b2 Ng8-f6 c2-c4 c7-c5 e2-e3 Nf8-g6 h2-h4 c5xb4 Bb2xf6 e7xf6 d2-d4 O-O
13/19 00:00 142k 959k -1.50 b2-b4 b7-b6 Bc1-b2 Ng8-f6 c2-c4 c7-c5 e2-e3 e7-e5 a2-a3 Nf8-e6 d2-d3 d7-d5 c4xd5 Nf6xd5
14/18 00:00 174k 971k -1.31 b2-b4 b7-b6 Bc1-b2 Ng8-f6 e2-e3 c7-c5 b4xc5 b6xc5 Ng1-f3 Nf8-g6 h2-h4 h7-h5 c2-c4 e7-e5 Nb1-c3 O-O
15/20 00:00 282k 1,006k -1.37 b2-b4 b7-b6 Ng1-f3 c7-c5 e2-e3 Ng8-f6 Bc1-b2 Nf8-g6 b4xc5 b6xc5 h2-h4 h7-h5 c2-c4 O-O Bb2xf6 g7xf6 Nb1-c3 Kg8-g7
16/22 00:00 333k 1,029k -1.29 b2-b4 Ng8-f6 Ng1-f3 b7-b6 Bc1-b2 Nf8-g6 e2-e3 O-O h2-h4 h7-h5 c2-c4 d7-d5 c4xd5 Ba8xd5 Nb1-c3
17/22 00:00 475k 1,014k -1.33 b2-b4 Ng8-f6 Bc1-b2 b7-b6 Ng1-f3 Nf8-g6 h2-h4 h7-h5 e2-e3 c7-c5 b4xc5 b6xc5 c2-c4 e7-e6 Nf3-g5 O-O Bb2xf6 g7xf6 Qd1xh5 f6xg5 h4xg5
18/22 00:00 729k 1,036k -1.36 b2-b4 e7-e5 Bc1-b2 Nf8-g6 h2-h4 h7-h5 e2-e3 d7-d5 g2-g3 Qc8-e6 Bf1-h3 Qe6-e7 d2-d4 Ng8-f6 d4xe5 Ng6xe5 Nb1-d2
19/25 00:01 1,786k 1,041k -0.94 d2-d4 d7-d5 b2-b4 Ng8-f6 a2-a4 Nf8-g6 Ng1-f3 O-O Bc1-b2 Nf6-e4 e2-e3 e7-e5 Nf3xe5 Ng6xe5 d4xe5 c7-c5
20/26 00:02 2,545k 1,034k -1.02 d2-d4 d7-d5 b2-b4 Ng8-f6 a2-a4 Nf8-g6 Ng1-f3 O-O e2-e3 Rf8-e8 Nb1-d2 b7-b6 b4-b5 c7-c5 b5xc6/ep Ba8xc6 Bc1-b2 e7-e5 d4xe5
21/26+ 00:03 3,567k 1,034k -0.94 d2-d4
21/26 00:04 4,161k 1,031k -0.99 d2-d4 Ng8-f6 c2-c4 b7-b5 c4xb5 c7-c5 d4xc5 Qc8xc5 e2-e3 Nf8-g6 Nb1-d2 O-O a2-a4 d7-d5 b2-b4 Qc5xb4 Bc1-a3 Qb4-a5 Ra1-c1
22/25- 00:04 4,353k 1,032k -1.07 d2-d4 Ng8-f6
22/25- 00:04 4,730k 1,028k -1.15 d2-d4 Ng8-f6
22/31+ 00:05 5,409k 1,021k -1.07 d2-d4
22/31- 00:05 5,644k 1,023k -1.27 d2-d4 Ng8-f6
22/31+ 00:05 5,772k 1,023k -1.08 d2-d4
22/31 00:06 6,168k 1,023k -1.01 d2-d4 Ng8-f6 c2-c4 c7-c5 d4-d5 b7-b5 Nb1-c3 b5-b4 Nc3-a4 Nf8-g6 Ng1-f3 O-O a2-a3 Rf8-e8 Bc1-g5 Nf6-e4 a3xb4 c5xb4 Ra1-c1 Kg8-h8 Bg5-e3 e7-e6 g2-g3
23/32+ 00:07 7,463k 1,014k -0.93 d2-d4
23/32- 00:07 8,012k 1,015k -1.09 d2-d4 Ng8-f6
23/32+ 00:08 8,175k 1,013k -0.97 d2-d4
23/32 00:08 8,468k 1,013k -1.06 d2-d4 Ng8-f6 c2-c4 b7-b6 Ng1-f3 d7-d5 c4xd5 Ba8xd5 Nb1-c3 Bd5-b7 Bc1-g5 c7-c5 Bg5xf6 g7xf6 d4-d5 Nf8-g6 Qd1-c2 Ke8-f8 e2-e3 Bb7xd5 Nc3xd5 Rd8xd5 Qc2-e4 Rd5-d8 Bf1-c4
24/35- 00:10 10,182k 1,014k -1.14 d2-d4 Ng8-f6
24/35+ 00:11 11,241k 1,010k -1.06 d2-d4
24/35+ 00:11 11,717k 1,012k -0.94 d2-d4
24/35 00:11 11,814k 1,012k -0.96 d2-d4 Ng8-f6 c2-c4 b7-b6 Ng1-f3 d7-d5 c4xd5 Ba8xd5 Nb1-c3 Bd5xf3 e2xf3 Nf8-e6 Bc1-e3 c7-c6 Bf1-c4 O-O Bc4xe6 Qc8xe6 Qd1-a4 Nf6-d5 Nc3xd5 Rd8xd5 Ra1-c1 Rd5-a5 Qa4xc6 Qe6xc6 Rc1xc6
25/31- 00:14 15,001k 1,007k -1.04 d2-d4 c7-c5
25/31+ 00:15 15,923k 1,002k -0.96 d2-d4
25/31 00:16 16,952k 1,002k -0.94 d2-d4 Ng8-f6 c2-c4 b7-b6 Ng1-f3 d7-d5 c4xd5 Ba8xd5 Nb1-c3 Bd5xf3 e2xf3 Nf8-e6 Bc1-e3 O-O Qd1-a4 c7-c5 d4xc5 Ne6xc5 Qa4-c2 Bb8-e5 Bf1-c4 Be5xc3+ Qc2xc3 Nc5-d3+ Bc4xd3 Qc8xc3+ b2xc3 Rd8xd3 Be3-d4
26/33- 00:19 19,726k 995k -1.02 d2-d4 Ng8-f6
26/33- 00:21 21,099k 995k -1.11 d2-d4 Ng8-f6
26/37- 00:33 33,029k 991k -1.23 d2-d4 Ng8-f6
26/42+ 00:35 35,608k 991k -1.12 d2-d4
26/42+ 00:38 37,838k 991k -0.88 d2-d4
26/42 00:38 38,515k 991k -1.04 d2-d4 Ng8-f6 Ng1-f3 c7-c5 d4-d5 b7-b5 c2-c4 e7-e6 Bc1-g5 b5xc4 Bg5xf6 g7xf6 Nb1-c3 Qc8-b7 e2-e4

[Chessqueen]

Here on the previous position I selected a Position for Black Pieces that give Dragon a Chcess960 position with over 1.04 advantage but it does not beat Stockfish 14 using the standard chess position, but when you give the same position to Stockfish 14 with Black Pieces after I tested 10 times Stockfish always beat Dragon 2.5.1. Remember the Black position is selected from Chess960 which give Black pieces an advantage of over 1.04

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.10.31"]
[Round "?"]
[White "Stockfish_14_x64_bmi2"]
[Black "Dragon-2.5-64bit-avx2"]
[Result "1/2-1/2"]
[BlackElo "3600"]
[Time "00:22:47"]
[WhiteElo "3640"]
[TimeControl "3600"]
[SetUp "1"]
[FEN "bbqrknnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "163"]
[WhiteType "program"]
[BlackType "program"]

1. b4 Nf6 2. Bb2 b6 3. c4 Ng6 4. h4 c5 5. bxc5 Ne5 6. cxb6 Nxc4 7. bxa7
Nxb2 8. axb8=Q Qxb8 9. Qc2 Rc8 10. Nc3 Nc4 11. e3 Na3 12. Qc1 Qb4 13. f3
O-O 14. Bd3 d5 15. Nge2 e5 16. Qd1 e4 17. Bc2 exf3 18. gxf3 Rc4 19. Rc1
Nxc2+ 20. Qxc2 d4 21. Nxd4 Rxd4 22. exd4 Bxf3 23. Rh3 Bg4 24. Re3 Qd6 25.
Ne4 Qh2 26. d3 Qh1+ 27. Kd2 Qh2+ 28. Ke1 Qg1+ 29. Kd2 Nxe4+ 30. Rxe4 Qh2+
31. Ke1 Qh1+ 32. Kd2 Qxh4 33. Rce1 Qg5+ 34. R1e3 Bd7 35. Qc5 Qg2+ 36. Re2
Qg1 37. Re1 Qh2+ 38. R1e2 Qh6+ 39. Ke1 Qh1+ 40. Kf2 Qh2+ 41. Kf3 Qh3+ 42.
Kf2 Qxd3 43. R4e3 Qa6 44. Qe5 Bg4 45. Rd2 h5 46. d5 Qh6 47. d6 Bd7 48. Rd5
Qg6 49. Qf4 Rc8 50. Re2 Bh3 51. Qf3 Bd7 52. Qf4 Bh3 53. Rg5 Qb1 54. Rg1 Qb7
55. Qe4 Qa7+ 56. Qe3 Qd7 57. Rd2 Bg4 58. Re1 Qb5 59. Qd4 Qg5 60. Qe3 Qa5
61. Qf4 Rc4 62. Re8+ Kh7 63. Re4 Rc3 64. Red4 Qd8 65. d7 Qh4+ 66. Kg1 Qe1+
67. Kg2 Qh4 68. Qf5+ Kh6 69. Qf4+ Kh7 70. d8=Q Qh3+ 71. Kg1 Rg3+ 72. Qxg3
Qxg3+ 73. Rg2 Qe1+ 74. Kh2 Qe5+ 75. Kg1 Qe1+ 76. Kh2 Qe5+ 77. Rg3 Qe2+ 78.
Kg1 Qe1+ 79. Kh2 Qe2+ 80. Kg1 Qe1+ 81. Kg2 Qe2+ 82. Kg1 {3-fold repetition}
1/2-1/2[/pgn]

[Chessqueen]

After the 11 game or game number #11 Komodo Dragon 2.5.1 got an even position after reaching this position, it is almost unbelievable to human Eyes that White can draw this position

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.10.31"]
[Round "11"]
[White "Dragon-2.5-64bit-avx2"]
[Black "Stockfish_14_x64_bmi2"]
[Result "1/2"]
[BlackElo "3640"]
[Time "07:33:12"]
[WhiteElo "3600"]
[TimeControl "3600"]
[SetUp "1"]
[FEN "bbqrknnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[Termination "unterminated"]
[PlyCount "108"]
[WhiteType "program"]
[BlackType "program"]


1. b4 Nf6 2. Bb2 Ng6 3. e3 b6 4. h4 h5 5. c4 d5 6. c5 e5 7. d4 O-O 8. Be2
bxc5 9. bxc5 Ng4 10. Bxg4 hxg4 11. h5 Ne7 12. dxe5 d4 13. Bxd4 Nf5 14. Nc3
Bxg2 15. Rh2 Bb7 16. h6 g6 17. Rb1 c6 18. f4 Nxd4 19. exd4 Ba6 20. Rg2 Bc7
21. Rb3 Qf5 22. Qxg4 Qxg4 23. Rxg4 Bc8 24. Rg3 Rxd4 25. Nge2 Rc4 26. Ra3
Bd8 27. Rd3 Bf5 28. Rd6 Rxc5 29. Nd4 Be7 30. Nxf5 gxf5 31. Rd3 Rxe5+ 32.
fxe5 Bxa3 33. Nd5 cxd5 34. Rxa3 Re8 35. Rxa7 Rxe5+ 36. Kf2 Re6 37. Rd7 Rxh6
38. Rxd5 Rh2+ 39. Kf3 Rxa2 40. Rd6 Kg7 41. Kf4 Rf2+ 42. Ke3 Rf1 43. Ra6
Re1+ 44. Kf3 Re6 45. Ra8 Kf6 46. Ra7 Rd6 47. Ra8 Rd4 48. Ke3 Rh4 49. Kf3
Rh3+ 50. Kf4 Rh4+ 51. Ke3 Rh2 52. Kf4 Rh4+ 53. Ke3 Re4+ 54. Kf3 Re1 55. Kf2
Re5 56. Kf3 Re6 57. Ra5 Re4 58. Ra7 Re7 59. Ra8 Re6 {3-fold repetition}
1/2-1/2[/pgn]

[Chessqueen]

After the 11 game or game number #11 Komodo Dragon 2.5.1 got an even position after reaching this position, it is almost unbelievable to human Eyes that White can draw this position

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.10.31"]
[Round "11"]
[White "Dragon-2.5-64bit-avx2"]
[Black "Stockfish_14_x64_bmi2"]
[Result "1/2"]
[BlackElo "3640"]
[Time "07:33:12"]
[WhiteElo "3600"]
[TimeControl "3600"]
[SetUp "1"]
[FEN "bbqrknnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[Termination "unterminated"]
[PlyCount "108"]
[WhiteType "program"]
[BlackType "program"]


1. b4 Nf6 2. Bb2 Ng6 3. e3 b6 4. h4 h5 5. c4 d5 6. c5 e5 7. d4 O-O 8. Be2
bxc5 9. bxc5 Ng4 10. Bxg4 hxg4 11. h5 Ne7 12. dxe5 d4 13. Bxd4 Nf5 14. Nc3
Bxg2 15. Rh2 Bb7 16. h6 g6 17. Rb1 c6 18. f4 Nxd4 19. exd4 Ba6 20. Rg2 Bc7
21. Rb3 Qf5 22. Qxg4 Qxg4 23. Rxg4 Bc8 24. Rg3 Rxd4 25. Nge2 Rc4 26. Ra3
Bd8 27. Rd3 Bf5 28. Rd6 Rxc5 29. Nd4 Be7 30. Nxf5 gxf5 31. Rd3 Rxe5+ 32.
fxe5 Bxa3 33. Nd5 cxd5 34. Rxa3 Re8 35. Rxa7 Rxe5+ 36. Kf2 Re6 37. Rd7 Rxh6
38. Rxd5 Rh2+ 39. Kf3 Rxa2 40. Rd6 Kg7 41. Kf4 Rf2+ 42. Ke3 Rf1 43. Ra6
Re1+ 44. Kf3 Re6 45. Ra8 Kf6 46. Ra7 Rd6 47. Ra8 Rd4 48. Ke3 Rh4 49. Kf3
Rh3+ 50. Kf4 Rh4+ 51. Ke3 Rh2 52. Kf4 Rh4+ 53. Ke3 Re4+ 54. Kf3 Re1 55. Kf2
Re5 56. Kf3 Re6 57. Ra5 Re4 58. Ra7 Re7 59. Ra8 Re6 {3-fold repetition}
1/2-1/2[/pgn]

Komodo Dragon had a bad start, but now after 36 games playing with standard position as White and Stockfish with the Black position that you see here now the score is 19 to to 17 in favor of Stockfish 14, now I am starting to see the reason why 1000 games are needed

[Chessqueen]

Revisiting my thoughts on this match, my opinion has now changed given that the match does not start until Nov 6th and that the GM does have a version of Komodo to practice with, I now think he should have good chances to win the match. I don't believe I gave enough thought to this little detail.

Since he is practicing with Komodo Dragon 2.5.1, he only need ChrisW opening book and Mr. Kaufman 32 core Threadripper, but I strongly recommend Mr. Kaufman to set his Komodo Dragon xxx.x to play at (5'+2") instead of (3' + 2")

I've run enough simulations to say that using more than 16 threads or more than 3' +2" (with 16 threads) is totally useless for knight odds. Imagine if we had a 32 man tablebase (from a million years in the future!). How would that help us in a knight odds match? Sure, if we got down to less than a pawn deficit in a game it might see a weird draw that normal search can't find, but against a knight-odds human opponent there is no point in assuming that he will play perfectly, it's better not to see every little detail in the far future that he will never even think of. If we get within drawing range we will surely not lose even if we play imperfectly, and if we reach a position where a win is possible we will probably win even if we miss the first win. Knight odds chess requires a different mental model than normal chess. Also, the ChrisW opening book is of no use for a human preparing, it is only for getting variety for computer vs computer matches. The positions (mostly) won't actually occur in real knight odds games.

If that is the case how come you will use a 32 core Threadripper instead of a 16 threads >

If we go with 16 threads then indeed I could use my fast 16 core machine instead. But everything is set up on my threadripper now, so there is no advantage to switching, and I might decide to use maybe 24 threads, depending on testing.

By now you probably have done several testing with 16 threads and with 24 threads; which seems to be more effective at Knight Odds against an engine of similar strength Gaviota-win32-0.84 = GM Perelshteyn ?

So far it seems 16 threads is best for this, but there are many ways to limit the search, by time, by depth, by Skill level, or by threads. So it's not a simple question, I'm trying to find the best combination. I'm hoping that we can draw the match, but I'm not predicting that.

So in the best case scenario all that you are hoping for is to get a draw with Knight Odds against a 2500+ rated player https://www.chess.com/players/eugene-perelshteyn

[lkaufman]

Revisiting my thoughts on this match, my opinion has now changed given that the match does not start until Nov 6th and that the GM does have a version of Komodo to practice with, I now think he should have good chances to win the match. I don't believe I gave enough thought to this little detail.

Since he is practicing with Komodo Dragon 2.5.1, he only need ChrisW opening book and Mr. Kaufman 32 core Threadripper, but I strongly recommend Mr. Kaufman to set his Komodo Dragon xxx.x to play at (5'+2") instead of (3' + 2")

I've run enough simulations to say that using more than 16 threads or more than 3' +2" (with 16 threads) is totally useless for knight odds. Imagine if we had a 32 man tablebase (from a million years in the future!). How would that help us in a knight odds match? Sure, if we got down to less than a pawn deficit in a game it might see a weird draw that normal search can't find, but against a knight-odds human opponent there is no point in assuming that he will play perfectly, it's better not to see every little detail in the far future that he will never even think of. If we get within drawing range we will surely not lose even if we play imperfectly, and if we reach a position where a win is possible we will probably win even if we miss the first win. Knight odds chess requires a different mental model than normal chess. Also, the ChrisW opening book is of no use for a human preparing, it is only for getting variety for computer vs computer matches. The positions (mostly) won't actually occur in real knight odds games.

If that is the case how come you will use a 32 core Threadripper instead of a 16 threads >

If we go with 16 threads then indeed I could use my fast 16 core machine instead. But everything is set up on my threadripper now, so there is no advantage to switching, and I might decide to use maybe 24 threads, depending on testing.

By now you probably have done several testing with 16 threads and with 24 threads; which seems to be more effective at Knight Odds against an engine of similar strength Gaviota-win32-0.84 = GM Perelshteyn ?

So far it seems 16 threads is best for this, but there are many ways to limit the search, by time, by depth, by Skill level, or by threads. So it's not a simple question, I'm trying to find the best combination. I'm hoping that we can draw the match, but I'm not predicting that.

So in the best case scenario all that you are hoping for is to get a draw with Knight Odds against a 2500+ rated player https://www.chess.com/players/eugene-perelshteyn

You mean a drawn match, not a draw. Of course I hope to win the match, just as I hoped to win the World Senior Championship in 2008, but didn't expect to win it. But I did win it, and it's possible we might win this match; it just wouldn't be a good bet in my opinion.

[Chessqueen]

Revisiting my thoughts on this match, my opinion has now changed given that the match does not start until Nov 6th and that the GM does have a version of Komodo to practice with, I now think he should have good chances to win the match. I don't believe I gave enough thought to this little detail.

Since he is practicing with Komodo Dragon 2.5.1, he only need ChrisW opening book and Mr. Kaufman 32 core Threadripper, but I strongly recommend Mr. Kaufman to set his Komodo Dragon xxx.x to play at (5'+2") instead of (3' + 2")

I've run enough simulations to say that using more than 16 threads or more than 3' +2" (with 16 threads) is totally useless for knight odds. Imagine if we had a 32 man tablebase (from a million years in the future!). How would that help us in a knight odds match? Sure, if we got down to less than a pawn deficit in a game it might see a weird draw that normal search can't find, but against a knight-odds human opponent there is no point in assuming that he will play perfectly, it's better not to see every little detail in the far future that he will never even think of. If we get within drawing range we will surely not lose even if we play imperfectly, and if we reach a position where a win is possible we will probably win even if we miss the first win. Knight odds chess requires a different mental model than normal chess. Also, the ChrisW opening book is of no use for a human preparing, it is only for getting variety for computer vs computer matches. The positions (mostly) won't actually occur in real knight odds games.

If that is the case how come you will use a 32 core Threadripper instead of a 16 threads >

If we go with 16 threads then indeed I could use my fast 16 core machine instead. But everything is set up on my threadripper now, so there is no advantage to switching, and I might decide to use maybe 24 threads, depending on testing.

By now you probably have done several testing with 16 threads and with 24 threads; which seems to be more effective at Knight Odds against an engine of similar strength Gaviota-win32-0.84 = GM Perelshteyn ?

So far it seems 16 threads is best for this, but there are many ways to limit the search, by time, by depth, by Skill level, or by threads. So it's not a simple question, I'm trying to find the best combination. I'm hoping that we can draw the match, but I'm not predicting that.

So in the best case scenario all that you are hoping for is to get a draw with Knight Odds against a 2500+ rated player https://www.chess.com/players/eugene-perelshteyn

You mean a drawn match, not a draw. Of course I hope to win the match, just as I hoped to win the World Senior Championship in 2008, but didn't expect to win it. But I did win it, and it's possible we might win this match; it just wouldn't be a good bet in my opinion.

Why did you decided to let him use the latest release of Komodo Dragon, unless you have a much better version that plays better against human with Knight Odds ?

[Chessqueen]

Revisiting my thoughts on this match, my opinion has now changed given that the match does not start until Nov 6th and that the GM does have a version of Komodo to practice with, I now think he should have good chances to win the match. I don't believe I gave enough thought to this little detail.

Since he is practicing with Komodo Dragon 2.5.1, he only need ChrisW opening book and Mr. Kaufman 32 core Threadripper, but I strongly recommend Mr. Kaufman to set his Komodo Dragon xxx.x to play at (5'+2") instead of (3' + 2")

I've run enough simulations to say that using more than 16 threads or more than 3' +2" (with 16 threads) is totally useless for knight odds. Imagine if we had a 32 man tablebase (from a million years in the future!). How would that help us in a knight odds match? Sure, if we got down to less than a pawn deficit in a game it might see a weird draw that normal search can't find, but against a knight-odds human opponent there is no point in assuming that he will play perfectly, it's better not to see every little detail in the far future that he will never even think of. If we get within drawing range we will surely not lose even if we play imperfectly, and if we reach a position where a win is possible we will probably win even if we miss the first win. Knight odds chess requires a different mental model than normal chess. Also, the ChrisW opening book is of no use for a human preparing, it is only for getting variety for computer vs computer matches. The positions (mostly) won't actually occur in real knight odds games.

If that is the case how come you will use a 32 core Threadripper instead of a 16 threads >

If we go with 16 threads then indeed I could use my fast 16 core machine instead. But everything is set up on my threadripper now, so there is no advantage to switching, and I might decide to use maybe 24 threads, depending on testing.

By now you probably have done several testing with 16 threads and with 24 threads; which seems to be more effective at Knight Odds against an engine of similar strength Gaviota-win32-0.84 = GM Perelshteyn ?

So far it seems 16 threads is best for this, but there are many ways to limit the search, by time, by depth, by Skill level, or by threads. So it's not a simple question, I'm trying to find the best combination. I'm hoping that we can draw the match, but I'm not predicting that.

So in the best case scenario all that you are hoping for is to get a draw with Knight Odds against a 2500+ rated player https://www.chess.com/players/eugene-perelshteyn

You mean a drawn match, not a draw. Of course I hope to win the match, just as I hoped to win the World Senior Championship in 2008, but didn't expect to win it. But I did win it, and it's possible we might win this match; it just wouldn't be a good bet in my opinion.

Why did you decided to let him use the latest release of Komodo Dragon, unless you have a much better version that plays better against human with Knight Odds ? I still do NOT understand why Komodo Dragon does not play better with more than 16 threads with Knight Odds, Is there also a limit as far as performing better with more than 7 pieces endgame for future Knight Odds matches

[lkaufman]

Revisiting my thoughts on this match, my opinion has now changed given that the match does not start until Nov 6th and that the GM does have a version of Komodo to practice with, I now think he should have good chances to win the match. I don't believe I gave enough thought to this little detail.

Since he is practicing with Komodo Dragon 2.5.1, he only need ChrisW opening book and Mr. Kaufman 32 core Threadripper, but I strongly recommend Mr. Kaufman to set his Komodo Dragon xxx.x to play at (5'+2") instead of (3' + 2")

I've run enough simulations to say that using more than 16 threads or more than 3' +2" (with 16 threads) is totally useless for knight odds. Imagine if we had a 32 man tablebase (from a million years in the future!). How would that help us in a knight odds match? Sure, if we got down to less than a pawn deficit in a game it might see a weird draw that normal search can't find, but against a knight-odds human opponent there is no point in assuming that he will play perfectly, it's better not to see every little detail in the far future that he will never even think of. If we get within drawing range we will surely not lose even if we play imperfectly, and if we reach a position where a win is possible we will probably win even if we miss the first win. Knight odds chess requires a different mental model than normal chess. Also, the ChrisW opening book is of no use for a human preparing, it is only for getting variety for computer vs computer matches. The positions (mostly) won't actually occur in real knight odds games.

If that is the case how come you will use a 32 core Threadripper instead of a 16 threads >

If we go with 16 threads then indeed I could use my fast 16 core machine instead. But everything is set up on my threadripper now, so there is no advantage to switching, and I might decide to use maybe 24 threads, depending on testing.

By now you probably have done several testing with 16 threads and with 24 threads; which seems to be more effective at Knight Odds against an engine of similar strength Gaviota-win32-0.84 = GM Perelshteyn ?

So far it seems 16 threads is best for this, but there are many ways to limit the search, by time, by depth, by Skill level, or by threads. So it's not a simple question, I'm trying to find the best combination. I'm hoping that we can draw the match, but I'm not predicting that.

So in the best case scenario all that you are hoping for is to get a draw with Knight Odds against a 2500+ rated player https://www.chess.com/players/eugene-perelshteyn

You mean a drawn match, not a draw. Of course I hope to win the match, just as I hoped to win the World Senior Championship in 2008, but didn't expect to win it. But I did win it, and it's possible we might win this match; it just wouldn't be a good bet in my opinion.

Why did you decided to let him use the latest release of Komodo Dragon, unless you have a much better version that plays better against human with Knight Odds ? I still do NOT understand why Komodo Dragon does not play better with more than 16 threads with Knight Odds, Is there also a limit as far as performing better with more than 7 pieces endgame for future Knight Odds matches

We just gave him Dragon 2.5.1, which he could buy anyway. We do have a better version, but not much better. With normal settings, in standard mode, Dragon won't do any better with big tablebases or more cores or more time. With MCTS mode, it does better with more time or more threads, but I'm not sure whether the time limit we are giving ourselves for this match is long enough for MCTS to do better than standard mode. However, I'm still trying to find ways to utilize the full power of the machine for this match, and I'm reasonably hopeful that I'll find a better solution than just letting it run in normal mode with 16 threads.