Skip to content

Instantly share code, notes, and snippets.

@nelhage
Created June 2, 2016 07:28
Show Gist options
  • Save nelhage/109d209a184a12efdb71183db0733230 to your computer and use it in GitHub Desktop.
Save nelhage/109d209a184a12efdb71183db0733230 to your computer and use it in GitHub Desktop.
2016/06/02 07:26:43 their-move game-id=44170 ply=21 ptn=11.B move="Sa1"
2016/06/02 07:26:43 seed=1464852403
2016/06/02 07:26:43 [minimax] deepen: depth=1 val=1495 pv=[5b1>14] time=116.247µs total=116.736µs evaluated=71 tt=0 branch=71
2016/06/02 07:26:43 [minimax] stats: visited=1 scout=0 evaluated=71 null=0/0 cut=0 cut0=0(0.00) cut1=0(0.00) m/cut=0.00 m/ms=619.370822 all=0 research=2
2016/06/02 07:26:43 [minimax] deepen: depth=2 val=1475 pv=[5b1>14 c2] time=214.212µs total=381.572µs evaluated=109 tt=1 branch=1
2016/06/02 07:26:43 [minimax] stats: visited=70 scout=68 evaluated=109 null=0/0 cut=67 cut0=67(0.99) cut1=0(0.99) m/cut=0.00 m/ms=835.620787 all=0 research=2
2016/06/02 07:26:43 [minimax] deepen: depth=3 val=1575 pv=[5b1>14 c2 e1] time=4.828232ms total=5.239369ms evaluated=4445 tt=9 branch=40
2016/06/02 07:26:43 [minimax] stats: visited=184 scout=181 evaluated=4445 null=0/0 cut=105 cut0=83(0.78) cut1=14(0.92) m/cut=9.67 m/ms=958.736034 all=67 research=2
2016/06/02 07:26:43 [minimax] deepen: depth=4 val=1430 pv=[a3 b2+ 5b1>14 c5] time=24.218742ms total=29.497973ms evaluated=10231 tt=284 branch=2
2016/06/02 07:26:43 [minimax] stats: visited=2203 scout=2192 evaluated=10231 null=54/67 cut=1632 cut0=1381(0.85) cut1=94(0.90) m/cut=8.01 m/ms=513.404041 all=222 research=8
2016/06/02 07:26:43 [minimax] deepen: depth=5 val=1630 pv=[a3 b2+ 5b1>14 c5 b2] time=109.130817ms total=138.66914ms evaluated=128291 tt=734 branch=12
2016/06/02 07:26:43 [minimax] stats: visited=5699 scout=5694 evaluated=128291 null=37/138 cut=2827 cut0=2475(0.88) cut1=143(0.93) m/cut=11.75 m/ms=1227.792513 all=2096 research=2
2016/06/02 07:26:43 [minimax] deepen: depth=6 val=1335 pv=[a3 a1+ 5b1>14 b2- e1 2b1>] time=165.743839ms total=304.439596ms evaluated=146221 tt=6695 branch=1
2016/06/02 07:26:43 [minimax] stats: visited=39215 scout=39183 evaluated=146221 null=2878/3489 cut=26273 cut0=24235(0.92) cut1=596(0.95) m/cut=9.11 m/ms=1118.810817 all=3338 research=34
2016/06/02 07:26:46 [minimax] deepen: depth=7 val=1455 pv=[a3 a1+ 5b1>14 b2- e1 2b1> b2] time=2.863182812s total=3.167679844s evaluated=4463609 tt=66643 branch=30
2016/06/02 07:26:46 [minimax] stats: visited=313892 scout=313885 evaluated=4463609 null=2660/8932 cut=173250 cut0=142307(0.82) cut1=6064(0.86) m/cut=14.61 m/ms=1668.597960 all=71332 research=1
2016/06/02 07:26:46 [minimax] time cutoff: depth=7 used=3.167679844s estimate=43.252973246s
@nelhage
Copy link
Author

nelhage commented Jun 2, 2016

breakdown:

deepen: depth=1 val=1495 pv=[5b1>14] time=116.247µs total=116.736µs evaluated=71 tt=0 branch=71

The bot just completed its search to depth=1 plies. The value of the terminal position was val=1495, and the sequence of moves it found was pv=[5b1>14]. This iteration took time=116.247µs and the total time spent so far this move was total=116.736µs. We evaluated evaluated=71 board positions as part of the search, and we short-cut tt=0 positions by returning pre-computed responses from the hash table. This iteration searched branch=71 times more positions than the previous iteration (at one ply shallower)

stats: visited=1 scout=0 evaluated=71 null=0/0 cut=0 cut0=0(0.00) cut1=0(0.00) m/cut=0.00 m/ms=619.370822 all=0 research=2

We visited visited=1 nodes, in which we searched some children. Of those we searched scout=0 in "scout" mode, which means we did a less-detailed search to try to prove that they weren't relevant, instead of doing a full search to find the best move. null=X/Y We employed the null move heuristic in Y positions, of which we pruned X . We searched m/ms=619.370822 positions per millisecond.

cut/all/research are related to the alpha-beta search and likely not super relevant.

I could easily export other information --- what might be fun to visualize?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment