Complicating code in C#

[Henk]

Also waste of resources.

Code: Select all

 32      15       216032      187713121    d2d4  d7d5  c1f4  c8f5  e2e3  e7e6  f1d3  f5d3  c2d3  f8d6  g2g3  g8e7  g1f3  b8d7  f4d6  c7d6  b1c3  e8g8  e1g1  e6e5  e3e4  e5d4  f3d4  d7f6  d4f3  g8h8  c3d5  e7d5  e4d5  f6d5  d3d4  a7a5

By the way Stockfish reaches level 33 in about two minutes on my computer. So really something wrong with Skipper. 2160/60 = 36 minutes.

[Henk]

Show must go on. Taking about ten minutes now.

Code: Select all

33       6        56852       49573610    e2e4  g8f6  b1c3  e7e5  g1f3  b8c6  f1c4  f8c5  d2d3  d7d6  c1e3  c8e6  e1g1  e8g8  c3d5  c6d4  a1c1  c7c6  d5f6  g7f6  c2c3  d4f3  g2f3  e6c4  d3c4  b7b5  c4b5  c6b5  f1e1  f8e8  f3f4  e8e6  e3c5  d6c5  

[Henk]

Yeah

Code: Select all

 
34       2        58137       53515983    e2e4  e7e5  c2c4  c7c5  f1d3  f8d6  b1c3  g8f6  g1f3  b8c6  e1g1  e8g8  b2b3  b7b6  c1b2  c8b7  c3d5  c6d4  b2c3  d4f3  d1f3  h7h5  f3d1  b7c6  a1c1  a7a6  d5f6  g7f6  f1e1  f8e8  g2g3  e8e6  a2a4  a8a7
35       4        72886       65592110    e2e4  e7e5  c2c4  c7c5  f1d3  f8d6  b1c3  g8f6  g1f3  b8c6  e1g1  e8g8  b2b3  b7b6  c1b2  c8b7  c3d5  c6d4  b2c3  d4f3  d1f3  h7h5  f3d1  b7c6  a1c1  a7a6  d5f6  g7f6  f1e1  f8e8  g2g3  b6b5  a2a4  b5c4  b3c4
36       2        76891       68920997    e2e4  e7e5  c2c4  c7c5  f1d3  f8d6  b1c3  g8f6  g1f3  b8c6  e1g1  e8g8  b2b3  b7b6  c1b2  c8b7  c3d5  c6d4  b2c3  d4f3  d1f3  h7h5  f3d1  b7c6  a1c1  a7a6  d5f6  g7f6  f1e1  a8a7  g2g3  g8h8  a2a3  f8e8  a3a4  e8e6

Ugly moves by the way. f1d3?

Stockfish only needs three minutes to produce a pv at level 36 on my machine. Skipper uses about 13 minutes while using dreadful LMR reductions. R = 20. So this result is a bad joke.

[Henk]

LMR= 20

Code: Select all

 
42      13        15995       15916096    d2d4  d7d5  c1f4  b8c6  e2e3  g8f6  g1f3  c8g4  f1d3  e7e6  h2h3  g4f5  d3f5  e6f5  b1c3  f8d6  f4g5  d6e7  e1g1  e8g8  a1b1  g8h8  h3h4  h7h6  g5f4  e7d6  f3e5  a7a5  f2f3  h8g8  g2g3  a8a7  a2a3  a5a4  g1h1  a7a5  f1g1  g8h8  h1h2  h8g8  g1h1  g8h8

LMR = 30

Code: Select all

 47      27        16926       14887659    g1f3  g8f6  d2d4  d7d5  e2e3  e7e6  f1d3  f8d6  b1c3  c7c5  c3b5  d6e7  c2c4  b8c6  e1g1  e8g8  c1d2  c8d7  d2c3  a8b8  g1h1  c5d4  e3d4  f8e8  f1g1  e7b4  a2a3  b4e7  g2g3  b7b6  a3a4  e7b4  g1f1  b4c3  b5c3  g7g5  f1e1  e8f8  h2h3  h7h5  a1a3  h5h4  c4d5  e6d5  e1g1  h4g3  f2g3

LMR = 40

Code: Select all

 52      -9        96304       79070457    e2e4  e7e5  g1f3  b8c6  d2d4  e5d4  f3d4  f8c5  c1e3  c6d4  e3d4  c5d4  d1d4  g8f6  c2c4  d7d6  b1c3  e8g8  d4d1  c8e6  f1d3  c7c6  e1g1  a8c8  a1c1  a7a6  h2h3  b7b5  b2b3  d6d5  e4d5  b5c4  b3c4  c6d5  c4d5  f6d5  c3e4  c8c1  d1c1  a6a5  f1d1  a5a4  f2f3  h7h6  g1h2  g8h8  a2a3  f8e8  d3c4  f7f6  d1h1  d5f4
 

[Henk]

LMR= 80. Maybe beter set it to 1000 and see what happens

Code: Select all

 82      24        39055       35257359    g1f3  d7d5  d2d4  g8f6  b1c3  e7e6  c1f4  f8d6  g2g3  b8c6  f1g2  e8g8  e1g1  a7a6  e2e3  d6f4  g3f4  c8d7  b2b3  a8c8  f1e1  f8e8  e3e4  d5e4  c3e4  f6d5  d1d2  b7b5  c2c3  g7g6  a1b1  e8f8  b1c1  f8e8  c1c2  e8f8  g1h1  f8e8  c2c1  h7h6  c1b1  e8f8  h1g1  f8e8  b1c1  e8f8  c1c2  f8e8  g1h1  g8h7  h2h4  h7h8  h1h2  h8g8  c2c1  e8f8  c1b1  f8e8  e1g1  c8b8  b1e1  b8c8  g1h1  e8f8  e1g1  f8e8  g1b1  g8h7  h1e1  h7g8
 83      25        55210       48781033    g1f3  d7d5  d2d4  g8f6  b1c3  e7e6  c1f4  f8d6  g2g3  b8c6  f1g2  e8g8  e1g1  a7a6  e2e3  d6f4  g3f4  c8d7  b2b3  a8c8  f1e1  b7b5  e3e4  f6e4  c3e4  d5e4  e1e4  b5b4  a1b1  g7g6  b1c1  f8e8  g1h1  h7h6  c1b1  e8f8  h1g1  f8e8  b1c1  e8f8  e4e1  f8e8  c1b1  g6g5  f4g5  h6g5  b1c1  e8f8  g1h1  f8e8  c1b1  e8f8  e1e3  f8e8  b1c1  e8f8  h1g1  f8e8  c1b1  e8f8  e3e4  f8e8  b1c1  e8f8  g1h1  f8e8  c1b1  e8f8  e4e1  f8e8  b1c1  e8f8  e1e3  f8e8  c1b1  e8f8  h1g1  f8e8  b1c1  e8f8  e3e1  f8e8  c1b1
 

[Sven]

Are you now solving chess?

[Henk]

No. It started with the problem why it was so difficult to search deeper than 20.

By the way taking too long now even with LMR=1000. Looks like it is getting stuck at level 90

Code: Select all

  90      19       161101      134697214    d2d4  g8f6  b1c3  d7d5  g1f3  e7e6  e2e3  c7c5  f1d3  b8c6  e1g1  c5d4  e3d4  f8d6  c3b5  d6b8  c1e3  e8g8  c2c4  c8d7  a2a3  d5c4  d3c4  a7a6  b5c3  b8d6  c4d3  f6d5  e3d2  f7f5  g2g3  b7b5  h2h4  h7h6  a1c1  h6h5  c1a1  f8e8  g1h1  e8f8  a1c1  f8e8  f1g1  e8f8  c1a1  a8a7  g1g2  a7c7  g2h2  g8h8  d2g5  d8c8  h2g2  f8e8  h1h2  h8g8  h2g1  e8f8  g1h1  f8e8  g2h2  g8h8  h2h3  h8g8  g5d2  c8d8  h3h2  g8h8  a1c1  h8g8  d2g5  d8c8  c1b1  e8f8  g5d2  c8d8  b1c1

[mvanthoor]

LMR= 80. Maybe beter set it to 1000 and see what happens

Set LMR to -50, so instead of reducing moves, it will extend them. It'll blow your mind.

Or, just write a new engine... that would probably waste much less of your time.

Wat is "level 90"? Is that 90 plies lookahead? Even Stockfish doesn't achieve that, except in pawn endgames. (But there, even my relatively primitive engine, already achieves 24 ply in under a second because of the transposition table.)

[Henk]

Level 90 means 90 plies distance to horizon. Looks like getting beyond 90 is a challenge. In end game it reaches level 450 . I saw when playing a game against an older version. By the way it lost the game with at least two blunders. Although coming back from a lost position. So it is playing unstable.

Demonstrates again that high search depth is no indication of strength. You can choose any depth you want. Although passing 90?

[Henk]

Skipper still too slow. When I set LMR = 1 I get no more than 60kn/s. So enough to do.