Next Komodo vs GM match.

[Vinvin]

Eval graph (1 second with latest SF) on Lichess : https://lichess.org/qED5HMmH#64
It seems 32...a6??? was the decisive mistake.

[mjlef]

Contempt of 150 or so is probably reasonable for knight odds matches, I think I used this value in some of them. But I'm pretty sure it's too high for a two pawn handicap game. The displayed evaluation of the initial handicap position should still be significantly negative or else the Contempt value is too high. The judgment is in the word "significantly".

Have you considering "Dynamic Contempt"? The idea here is that Komodo wants to believe that it has some advantage in the position, say, a 0.90 advantage, so, on the next move, if the previous one had 0.80 advantage, it raises contempt by 10, but if it was 1.00, it reduces it by 10.

The idea here is that Komodo adjusts its contempt depending on how the opponent plays, so that, if Komodo has already equalized, it now plays with Contempt 0 because it assumes it can beat the human from an equal position.

We actually do vary Contempt, not exactly like you state, but along lines like that. It is already little "dynamic". I like it when others come up with some of our ideas like this. It helps confirm if it could be good.

[lkaufman]

Eval graph (1 second with latest SF) on Lichess : https://lichess.org/qED5HMmH#64
It seems 32...a6??? was the decisive mistake.

Yes, it seems 32...b4 might still have held with perfect play.

[Vinvin]

Eval graph (1 second with latest SF) on Lichess : https://lichess.org/qED5HMmH#64
It seems 32...a6??? was the decisive mistake.

Yes, it seems 32...b4 might still have held with perfect play.

After a longer analyze, 31...b5 is already a blunder because of the played move 32.h5!! (+1 eval : 32. .. b4 33. Nf7+ Kg8 34. h6 Rc2 35. Re7 Nc4 36. Ne5 ...).
31...Nb3 or 31...Rd1+ with about equal eval.

[Vinvin]

...

Hello Larry,
a question about your new computer (16 core I-9) : how does it perform compare to the older computer ?

[lkaufman]

Izoria was winning the endgame with two extra pawns but first blundered the win with 38.c3?? and then blew the draw with 40.Rxc5??. At the end people were still unsure who was winning, but Komodo announced forced mate before Izoria resigned.


[Event "Rapid 45m + 15s"]
[Site "chess.com"]
[Date "2018.01.15"]
[Round "2"]
[White "Izoria, Zviad"]
[Black "Komodo 2000.00 64-bit"]
[Result "0-1"]
[Annotator "Kaufman,Larry"]
[SetUp "1"]
[FEN "rnbqkbnr/p1pppp1p/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "84"]
[TimeControl "2100+8"]

{6144MB, Book2016.ctg, DESKTOP-16c} 1. e4 {0} c5 {1.10/29 37} 2. Nf3 {0} Bb7 {
1.17/30 34} 3. Nc3 {38} Nf6 {1.22/31 32} 4. e5 {138} Ne4 {1.41/32 0} 5. Bd3 {
150} Nxc3 {1.48/33 0} 6. dxc3 {54} Nc6 {1.50/32 36} 7. O-O {25} Qc7 {1.56/31 72
} 8. Qe2 {116} e6 {1.55/30 68} 9. Bf4 {86} Ne7 {1.42/30 63} 10. Be4 {236} Ng6 {
1.33/30 46} 11. Bg3 {40} Bxe4 {1.49/34 18} 12. Qxe4 {23} Rb8 {1.49/34 9} 13. b3
{27} h5 {1.63/35 36} 14. Rfd1 {250} Qb7 {1.47/34 37} 15. Qxb7 {48} Rxb7 {
1.41/30 19} 16. Nh4 {67} a5 {1.33/34 95} 17. c4 {229} a4 {1.47/39 0} 18. Rd3 {
70} Ne7 {1.28/34 60} 19. f4 {198} Nc6 {1.44/37 0} 20. Nf3 {153} Be7 {1.52/36 0}
21. Be1 {86} f6 {1.34/31 55} 22. exf6 {122} Bxf6 {1.47/35 0} 23. Rad1 {16} Rc7
{1.44/33 53} 24. f5 {158} Nd4 {1.51/36 0} 25. fxe6 {100} dxe6 {1.42/33 34} 26.
Nxd4 {61} Bxd4+ {1.45/36 0} 27. Bf2 {41} Bxf2+ {1.52/36 0} 28. Kxf2 {12} Ke7 {
1.61/38 77} 29. Rg3 {58} Rcc8 {1.46/33 27} 30. Rg7+ {182} Kf6 {1.50/40 0} 31.
Rdd7 {16} Rcd8 {1.35/35 34} 32. Rdf7+ {209} Ke5 {1.76/42 0} 33. Ke3 {127} Rhg8
{1.87/38 264} 34. g3 {55} Rxg7 {2.03/39 30} 35. Rxg7 {13} Rd1 {2.18/38 49} 36.
Rg5+ {162} Kd6 {2.55/42 0} 37. Rxh5 {16} a3 {1.55/33 17} 38. c3 {218} Ra1 {
0.00/37 19} 39. b4 {142} Rxa2 {0.00/42 23} 40. Rxc5 {25} Ra1 {-4.28/29 29} 41.
Ra5 {220} a2 {-7.26/31 0} 42. c5+ {20} Ke5 {-250.00/35 101} 0-1

[Vinvin]

Izoria was winning the endgame with two extra pawns but first blundered the win with 38.c3?? and then blew the draw with 40.Rxc5??. At the end people were still unsure who was winning, but Komodo announced forced mate before Izoria resigned.

...

Graph eval with SF, 1 second per move : https://lichess.org/1ebW7i8v#78

[Nordlandia]

Stockfish 130118 find 42...Ke5 suboptimal (still enough to win) it prefers Kd5 though. Btw SF consulted 6-men syzygy from NVMe SSD. Maybe new motherboard support NVMe SSD Larry?

[Vinvin]

Stockfish 130118 find 42...Ke5 suboptimal (still enough to win) it prefers Kd5 though. Btw SF consulted 6-men syzygy from NVMe SSD. Maybe new motherboard support NVMe SSD Larry?

It's a bit suboptimal but not that much :

Code: Select all

Stockfish_180113_x64_modern:
...
 42/44	00:32	 395 193k	12 206k	-132,61	42. ... Ke5 43.Ra7 Re1+ 44.Kd3 a1Q 45.Rxa1 Rxa1 46.Kc4 Ra2 47.h4 Ra8 48.b5 Ra4+ 49.Kb3 Rg4 50.c4 Kf6 51.b6 Ke7 52.Kb4 Kd7 53.Kb5 Rxg3 54.c6+ Kc8 
 42/34	00:32	 395 193k	12 206k	-132,65	42. ... Kd5 43.Ra7 Re1+ 44.Kd3 Rd1+ 45.Kc2 a1Q 46.Rxa1 Rxa1 47.Kd3 Rd1+ 48.Ke2 Rh1 49.Kd3 Rxh2 50.c4+ Kc6 51.g4 Rg2 52.g5 Rxg5 53.Kd2 Rg3 54.Kc2 

[lkaufman]

...

Hello Larry,
a question about your new computer (16 core I-9) : how does it perform compare to the older computer ?

For single core testing my new machine and my older 24 core are about equal because the i9 is about 3/2 times faster but 2/3 the cores. But for playing a single game on all cores the new one is clearly superior, since the benefit of 50% more cores is way less than the benefit from 50% more speed. Probably it's less than ten elo stronger at these longer time limit games.