An easy position, for humans

[peter]

This is 24 plies ahead, but Asmfish, again, can't see the draw. It quickly reaches depth 43, then it needs a lot of time for depth 44....

[d]8/8/1P1k4/3P2p1/4KpP1/8/8/8 b - - 0 25

That's even some more crass cause it's already a 7some drawn position.
Very good find again, Rodolfo, thanks!

Shredder does it again even without tbs.

8/8/1P1k4/3P2p1/4KpP1/8/8/8 b - - 0 1

Analysis by Deep Shredder 13 x64:
...
25...Kd7 26.Ke5 f3 27.b7 Kc7 28.Ke6 Kxb7 29.d6 f2 30.d7 f1Q 31.d8Q Qf4 32.Qd5+ Kc7 33.Qc5+ Kb7 34.Qf5
+- (2.33 ++) Depth: 45/83 00:00:20 958MN
...
25...Kd7 26.Ke5 f3 27.b7 Kc7 28.Ke6 Kxb7 29.d6 f2 30.d7 f1Q 31.d8Q Qf4 32.Qd5+ Kb8 33.Qf5 Kc8 34.Kf6+ Kd8 35.Kg6 Ke8 36.Qc8+ Ke7 37.Qc5+ Ke8 38.Qxg5 Qd6+ 39.Qf6 Qd3+ 40.Qf5 Qa6+ 41.Kg7 Qb7+ 42.Kh6 Qb2 43.Qe6+ Kf8 44.Qd6+ Kg8 45.Qd8+ Kf7 46.Qc7+ Kf8 47.Qc5+ Kg8 48.Qc8+ Kf7 49.Qd7+ Kf8 50.Qa7
+/- (1.17 ++) Depth: 45/83 00:00:25 1211MN
...
25...Kd7 26.Ke5 f3 27.b7 Kc7 28.Ke6 Kxb7 29.d6 f2 30.d7 f1Q 31.d8Q Qf4 32.Qd5+ Kc7 33.Qc5+ Kb8 34.Qf5 Kc8 35.Kf6+ Kd8 36.Kg6 Ke8 37.Qxg5 Qd6+ 38.Qf6 Qd3+ 39.Qf5 Qa6+ 40.Kh5 Qb7 41.Kh6 Qb2 42.Qg6+ Kf8 43.Qd6+ Kg8 44.Qd8+ Kf7 45.Qd7+ Kf8 46.g5 Qh2+ 47.Kg6 Qc2+ 48.Kf6 Qc3+ 49.Ke6 Qh3+ 50.Kd6 Qxd7+ 51.Kxd7 Kf7 52.g6+ Kxg6 53.Ke7 Kg7 54.Ke8 Kh8 55.Ke7 Kh7 56.Kf7 Kh6 57.Kf6 Kh7 58.Kg5 Kg7 59.Kh5 Kh8 60.Kh6 Kg8 61.Kg5 Kf7
= (0.00) Depth: 45/87 00:00:37 1725MN
...
25...Kd7 26.Ke5 f3 27.b7 Kc7 28.Ke6 Kxb7 29.d6 f2 30.d7 f1Q 31.d8Q Qf4 32.Qd3 Qxg4+ 33.Kf6 Qf4+ 34.Kg6 g4 35.Qb5+ Kc7 36.Qc5+ Kd7 37.Qb5+ Ke6 38.Qe8+ Kd5 39.Qc8 Kd4 40.Kh5 g3 41.Qc6 Ke5 42.Qc7+ Ke4 43.Qc6+ Kf5 44.Qg6+ Ke5 45.Qc6 Qe3 46.Kg4 Qf2 47.Qc3+ Ke6 48.Qxg3 Qxg3+ 49.Kxg3 Kd6 50.Kf2 Kd7 51.Ke3 Kc7 52.Kd3 Kd7 53.Ke2 Kd6 54.Kd3 Kd7
= (0.00) Depth: 60/96 00:03:53 11560MN[/quote]

[Rodolfo Leoni]

Depth 60.... Amazing!

Rodolfo

[Joerg Oster]

Not a really difficult position, isn't it?

1 Thread, 512 MB Hash, no EGTBs:

Code: Select all

info depth 19 seldepth 34 multipv 1 score cp -118 nodes 129496 nps 1027746 tbhits 0 time 126 pv d6d7 e4d4 d7d6 d4d3 d6d7 d3e4 d7d6 e4f3 d6d7 f3e2 d7d6 e2f2 d6d7 f2f3 d7d6 f3g2 d6d7 g2f2 d7d6 f2e2 d6d7 e2d3 d7d6 d3c3 d6d7 c3c4 d7c8 d5d6 f4f3
info depth 20 seldepth 33 multipv 1 score cp -57 nodes 194125 nps 1329623 tbhits 0 time 146 pv d6d7 e4d4 d7d6 d4d3 d6d7 d3e4 d7d6 e4f3 d6d7 f3e2 d7d6 e2f2 d6d7 f2f3 d7d6 f3g2 d6d7 g2f2 d7d6 f2e2 d6d7 e2d3 d7d6 d3c3 d6d7 c3c4 d7d6 c4d4 d6d7 d4e5 f4f3


info depth 43 seldepth 91 multipv 1 score cp -15 nodes 291314877 nps 3553357 hashfull 845 tbhits 0 time 81983 pv d6d7 e4e5 f4f3 b6b7 d7c7 d5d6 c7b7 e5e6 f3f2 d6d7 f2f1q d7d8q f1f4 d8d5 b7c7 d5a5 c7c8 a5f5 c8d8 e6f6 d8e8 f6g7 e8e7 f5c5 e7e6 c5c8 e6e5 c8e8 e5d6 e8e2 d6d7 g7g6 d7d6 e2a6 d6e7 a6b7 e7f8 b7c8 f8e7 c8c5 e7e8 c5g5 f4d6 g5f6 d6d3 f6f5 d3a6 g6g5 a6b7 g5h6 b7b2 h6g6 b2b6 g6h5 b6b7 h5g5 b7g7 g5h4 g7c3 f5g6 e8f8 g6d6 f8g8 d6e6 g8f8 g4g5 c3d4 h4h3 d4f4
info depth 44 currmove d6d7 currmovenumber 1
info depth 44 seldepth 94 multipv 1 score cp -8 lowerbound nodes 356430315 nps 3560671 hashfull 891 tbhits 0 time 100102 pv d6d7
info depth 44 currmove d6d7 currmovenumber 1
info depth 44 currmove f4f3 currmovenumber 2
info depth 44 seldepth 85 multipv 1 score cp -8 nodes 404171004 nps 3571077 hashfull 906 tbhits 0 time 113179 pv d6d7 e4e5 f4f3 b6b7 d7c7 d5d6 c7b7 e5e6 f3f2 d6d7 f2f1q d7d8q f1f4 d8d5 b7c7 d5a5 c7c8 a5c5 c8b8 c5b5 b8c7 b5e5 c7c8 e5h8 c8b7 h8h7 b7c6 h7f5 c6c7 e6f6 c7d8 f6g6 d8e8 g6g7 e8e7 f5c5 e7e6 c5c8 e6e5 c8e8 e5d6 e8d8 d6e6 d8b6 e6e5 b6b2 e5d6 b2e2 d6d7 g7g6 d7d6 e2a6 d6e7 a6b7 e7f8 b7c8 f8e7 c8c5 e7e8 c5g5 f4d6 g6g7 d6c7 g7h6 e8f8 g5f6 f8g8 f6g6 g8f8 g6f5 f8g8 f5d5 g8f8 g4g5 c7h2 h6g6 h2c2 g6f6

[peter]

Depth 60.... Amazing!

Well yes, but maybe it's rather the selective depth (96 here according to the 60 "nonselective") that's the miracle here.

This could be depth of "singular extensions" of course and maybe it's not even a matter of special Shredder- "knowledge" and hard- coded static eval here but the coming of 50moves boundary into extensions' horizon

[Rodolfo Leoni]

Not a really difficult position, isn't it?

1 Thread, 512 MB Hash, no EGTBs:

Code: Select all

info depth 19 seldepth 34 multipv 1 score cp -118 nodes 129496 nps 1027746 tbhits 0 time 126 pv d6d7 e4d4 d7d6 d4d3 d6d7 d3e4 d7d6 e4f3 d6d7 f3e2 d7d6 e2f2 d6d7 f2f3 d7d6 f3g2 d6d7 g2f2 d7d6 f2e2 d6d7 e2d3 d7d6 d3c3 d6d7 c3c4 d7c8 d5d6 f4f3
info depth 20 seldepth 33 multipv 1 score cp -57 nodes 194125 nps 1329623 tbhits 0 time 146 pv d6d7 e4d4 d7d6 d4d3 d6d7 d3e4 d7d6 e4f3 d6d7 f3e2 d7d6 e2f2 d6d7 f2f3 d7d6 f3g2 d6d7 g2f2 d7d6 f2e2 d6d7 e2d3 d7d6 d3c3 d6d7 c3c4 d7d6 c4d4 d6d7 d4e5 f4f3


info depth 43 seldepth 91 multipv 1 score cp -15 nodes 291314877 nps 3553357 hashfull 845 tbhits 0 time 81983 pv d6d7 e4e5 f4f3 b6b7 d7c7 d5d6 c7b7 e5e6 f3f2 d6d7 f2f1q d7d8q f1f4 d8d5 b7c7 d5a5 c7c8 a5f5 c8d8 e6f6 d8e8 f6g7 e8e7 f5c5 e7e6 c5c8 e6e5 c8e8 e5d6 e8e2 d6d7 g7g6 d7d6 e2a6 d6e7 a6b7 e7f8 b7c8 f8e7 c8c5 e7e8 c5g5 f4d6 g5f6 d6d3 f6f5 d3a6 g6g5 a6b7 g5h6 b7b2 h6g6 b2b6 g6h5 b6b7 h5g5 b7g7 g5h4 g7c3 f5g6 e8f8 g6d6 f8g8 d6e6 g8f8 g4g5 c3d4 h4h3 d4f4
info depth 44 currmove d6d7 currmovenumber 1
info depth 44 seldepth 94 multipv 1 score cp -8 lowerbound nodes 356430315 nps 3560671 hashfull 891 tbhits 0 time 100102 pv d6d7
info depth 44 currmove d6d7 currmovenumber 1
info depth 44 currmove f4f3 currmovenumber 2
info depth 44 seldepth 85 multipv 1 score cp -8 nodes 404171004 nps 3571077 hashfull 906 tbhits 0 time 113179 pv d6d7 e4e5 f4f3 b6b7 d7c7 d5d6 c7b7 e5e6 f3f2 d6d7 f2f1q d7d8q f1f4 d8d5 b7c7 d5a5 c7c8 a5c5 c8b8 c5b5 b8c7 b5e5 c7c8 e5h8 c8b7 h8h7 b7c6 h7f5 c6c7 e6f6 c7d8 f6g6 d8e8 g6g7 e8e7 f5c5 e7e6 c5c8 e6e5 c8e8 e5d6 e8d8 d6e6 d8b6 e6e5 b6b2 e5d6 b2e2 d6d7 g7g6 d7d6 e2a6 d6e7 a6b7 e7f8 b7c8 f8e7 c8c5 e7e8 c5g5 f4d6 g6g7 d6c7 g7h6 e8f8 g5f6 f8g8 f6g6 g8f8 g6f5 f8g8 f5d5 g8f8 g4g5 c7h2 h6g6 h2c2 g6f6

Hi Joerg.

I tried many engines, but none of them got that performance... What engine did you use? Is it available for download?

Thanks,
Rodolfo

[peter]

Not a really difficult position, isn't it?

1 Thread, 512 MB Hash, no EGTBs:

Of course you're right, Jörg, tbs should be turned off better, there aren't any 6some of interest to be reached anyhow.

Yet SF dev needs depth 52 for the eval to go beyond 2 at 24 threads and 32G hash:

8/8/1P1k4/3P2p1/4KpP1/8/8/8 b - - 0 1

Analysis by Stockfish 140717 64 POPCNT:
...
25...Kd7 26.Kd3 Kd6 27.Kd4 Kd7 28.Kc4 Kd6 29.Kc3 Kd7 30.Kc2 Kd6 31.Kd3 Kd7 32.Ke4 Kd6 33.Kf3 Kd7 34.Kf2 Kd6 35.Ke2 Kd7 36.Kf3 Kd6 37.Kf2 Kd7 38.Ke1 Kd6 39.Kf1 Kd7 40.Kg1 Kd6 41.Kh2 Kd7 42.Kg2 Kd6 43.Kh1 Kd7 44.Kh2 Kd6 45.Kg1 Kd7 46.Kh1 Kd6 47.Kg2 Kd7 48.Kf1 Kd6 49.Ke1 Kd7 50.Kd1 Kd6 51.Kc1 Kd7 52.Kd2 Kd6 53.Kd1 Kd7 54.Ke2 Kd6 55.Kd2 Kd7 56.Kc1 Kd6 57.Kc2 Kd7 58.Kc3 Kd6 59.Kc4 Kd7
+- (2.17) Depth: 51/74 00:00:33 1301MN
25...Kd7
+- (2.10 ++) Depth: 52/81 00:00:40 1651MN
25...Kd7
+- (2.03 ++) Depth: 52/81 00:00:41 1733MN
25...Kd7
+- (1.92 ++) Depth: 52/81 00:00:47 2026MN
...
25...Kd7
+/- (0.90 ++) Depth: 52/81 00:06:14 21654MN

Took me too long to wait for the ++ to be solved.

Very big SMP- pity at my try or a special SF- version of yours at your try


Edit: saw your answer to my question already in yours to Rodolfo.
What's the point of the new patch here

[Joerg Oster]

Not a really difficult position, isn't it?

1 Thread, 512 MB Hash, no EGTBs:

Code: Select all

info depth 19 seldepth 34 multipv 1 score cp -118 nodes 129496 nps 1027746 tbhits 0 time 126 pv d6d7 e4d4 d7d6 d4d3 d6d7 d3e4 d7d6 e4f3 d6d7 f3e2 d7d6 e2f2 d6d7 f2f3 d7d6 f3g2 d6d7 g2f2 d7d6 f2e2 d6d7 e2d3 d7d6 d3c3 d6d7 c3c4 d7c8 d5d6 f4f3
info depth 20 seldepth 33 multipv 1 score cp -57 nodes 194125 nps 1329623 tbhits 0 time 146 pv d6d7 e4d4 d7d6 d4d3 d6d7 d3e4 d7d6 e4f3 d6d7 f3e2 d7d6 e2f2 d6d7 f2f3 d7d6 f3g2 d6d7 g2f2 d7d6 f2e2 d6d7 e2d3 d7d6 d3c3 d6d7 c3c4 d7d6 c4d4 d6d7 d4e5 f4f3


info depth 43 seldepth 91 multipv 1 score cp -15 nodes 291314877 nps 3553357 hashfull 845 tbhits 0 time 81983 pv d6d7 e4e5 f4f3 b6b7 d7c7 d5d6 c7b7 e5e6 f3f2 d6d7 f2f1q d7d8q f1f4 d8d5 b7c7 d5a5 c7c8 a5f5 c8d8 e6f6 d8e8 f6g7 e8e7 f5c5 e7e6 c5c8 e6e5 c8e8 e5d6 e8e2 d6d7 g7g6 d7d6 e2a6 d6e7 a6b7 e7f8 b7c8 f8e7 c8c5 e7e8 c5g5 f4d6 g5f6 d6d3 f6f5 d3a6 g6g5 a6b7 g5h6 b7b2 h6g6 b2b6 g6h5 b6b7 h5g5 b7g7 g5h4 g7c3 f5g6 e8f8 g6d6 f8g8 d6e6 g8f8 g4g5 c3d4 h4h3 d4f4
info depth 44 currmove d6d7 currmovenumber 1
info depth 44 seldepth 94 multipv 1 score cp -8 lowerbound nodes 356430315 nps 3560671 hashfull 891 tbhits 0 time 100102 pv d6d7
info depth 44 currmove d6d7 currmovenumber 1
info depth 44 currmove f4f3 currmovenumber 2
info depth 44 seldepth 85 multipv 1 score cp -8 nodes 404171004 nps 3571077 hashfull 906 tbhits 0 time 113179 pv d6d7 e4e5 f4f3 b6b7 d7c7 d5d6 c7b7 e5e6 f3f2 d6d7 f2f1q d7d8q f1f4 d8d5 b7c7 d5a5 c7c8 a5c5 c8b8 c5b5 b8c7 b5e5 c7c8 e5h8 c8b7 h8h7 b7c6 h7f5 c6c7 e6f6 c7d8 f6g6 d8e8 g6g7 e8e7 f5c5 e7e6 c5c8 e6e5 c8e8 e5d6 e8d8 d6e6 d8b6 e6e5 b6b2 e5d6 b2e2 d6d7 g7g6 d7d6 e2a6 d6e7 a6b7 e7f8 b7c8 f8e7 c8c5 e7e8 c5g5 f4d6 g6g7 d6c7 g7h6 e8f8 g5f6 f8g8 f6g6 g8f8 g6f5 f8g8 f5d5 g8f8 g4g5 c7h2 h6g6 h2c2 g6f6

Hi Joerg.

I tried many engines, but none of them got that performance... What engine did you use? Is it available for download?

Thanks,
Rodolfo

As can be seen from the output it's a slightly modified Stockfish.
You can find its sources here https://github.com/joergoster/Stockfish/tree/zander

But the above output is from an experimental version which has not been pushed to the repository by now.

[Joerg Oster]

Not a really difficult position, isn't it?

1 Thread, 512 MB Hash, no EGTBs:

Of course you're right, Jörg, tbs should be turned off better, there aren't any 6some of interest to be reached anyhow.

Yet SF dev needs depth 52 for the eval to go beyond 2 at 24 threads and 32G hash:

8/8/1P1k4/3P2p1/4KpP1/8/8/8 b - - 0 1

Analysis by Stockfish 140717 64 POPCNT:
...
25...Kd7 26.Kd3 Kd6 27.Kd4 Kd7 28.Kc4 Kd6 29.Kc3 Kd7 30.Kc2 Kd6 31.Kd3 Kd7 32.Ke4 Kd6 33.Kf3 Kd7 34.Kf2 Kd6 35.Ke2 Kd7 36.Kf3 Kd6 37.Kf2 Kd7 38.Ke1 Kd6 39.Kf1 Kd7 40.Kg1 Kd6 41.Kh2 Kd7 42.Kg2 Kd6 43.Kh1 Kd7 44.Kh2 Kd6 45.Kg1 Kd7 46.Kh1 Kd6 47.Kg2 Kd7 48.Kf1 Kd6 49.Ke1 Kd7 50.Kd1 Kd6 51.Kc1 Kd7 52.Kd2 Kd6 53.Kd1 Kd7 54.Ke2 Kd6 55.Kd2 Kd7 56.Kc1 Kd6 57.Kc2 Kd7 58.Kc3 Kd6 59.Kc4 Kd7
+- (2.17) Depth: 51/74 00:00:33 1301MN
25...Kd7
+- (2.10 ++) Depth: 52/81 00:00:40 1651MN
25...Kd7
+- (2.03 ++) Depth: 52/81 00:00:41 1733MN
25...Kd7
+- (1.92 ++) Depth: 52/81 00:00:47 2026MN
...
25...Kd7
+/- (0.90 ++) Depth: 52/81 00:06:14 21654MN

Took me too long to wait for the ++ to be solved.

Very big SMP- pity at my try or a special SF- version of yours at your try

The latter.

[peter]

Cross- writing and -reading.
What's the difference as for that position and the new patch

[Lyudmil Tsvetkov]

time might have come to u to reinvent the whole game from the very beginning or even the whole planet is actually more sufficent maybe

I don't understand you.

if precisely 20 years ago the very top human player, Kasparov only managed to draw Deep Blue, roughly, and now SF is 700 elo stronger than Deep Blue/Kasparov, definitely a lot of water has passed under the chess bridge.

including under the theoretical chess bridge.

in 20 more years, current SF will be another 700 elo or so weaker than the then strongest engine, so definitely a whole lot more knowledge to uncover and strength to gain.

what is so unusual about that?

people nowadays, who would whole-heartedly want to do that, definitely process 10 times more information that Kasparov, and also much more valuable information, as considerable part of it might be checked with SF.

so, it is much easier nowadays to decide which opening moves are best than that was the case 20 years ago.

1.c4 is definitely best.

just checked last 2 remaining off-beat black replies, g6 and e6(f5 fails to 2.d4 e6 3.d5), and with very high certainty am able to claim 1.c4 should win in some 90% of lines, whatever that means.

the precentages for 1.e4 and 1.d4 are definitely much lower, especially for d4.

on 1.e4, black has c5, very close to equality, while on 1.d4 both the best c5, which leads to a draw some dozens moves later, and the Dutch stonewall with 1...d5 2.c4 e6 3.Nc3 f5, leading to a draw 30 moves later, equalise completely, with a known outcome.

I'd be glad if you stay out of topics I start.

As an exchange, I'll stay out of the topics you start. (BTW, I can't find any.)

is that down in the forum Charter somewhere?

that I should stay out of your topics?

anyway, I promise to stay out of any of your topics, concerning aliens, star ships, commanders Komodo and SF, etc.

if anyone breaks the charter, that is only you.

look at what you have been posting about, and then criticise others.