With big hash Natwarlal v0.06 searches pretty fast...

[Dann Corbit]

Geschrieben von:/Posted by: Dann Corbit at 24 September 2004 07:03:17:


Natwarlal v0.06
by Pallav Nawani, India.
Natwarlal is a winboard compatible chess engine. Natwarlal is released under
MIT license. See license.txt (included with this program) for details.
Type 'help' for a list of commands.
Initialized Hashtables.
Hashtable 131072K, Pawn hashtable 32768K
Opening book initialized (from natwarlal.bk).
Positions in book: 43579
Ready to roll! (PS: Type 'new' to start a game)
Natwarlal> st 999
Natwarlal> sd 999
Natwarlal> post
Natwarlal> new
Black
8 r n b q k b n r
7 p p p p p p p p
6 - - - - - - - -
5 - - - - - - - -
4 - - - - - - - -
3 - - - - - - - -
2 P P P P P P P P
1 R N B Q K B N R
a b c d e f g h
White
White to move [KQkq[-]
Natwarlal> go
Time alotted 3600s, ext 1200
1 0 0 21 d2d4 d7d5
1 0 0 65 d2d4 d7d5
2 25 0 141 d2d4 d7d5 e2e3
2 25 0 651 d2d4 d7d5 e2e3
3 0 0 807 d2d4 d7d5 e2e3 e7e6
Searching depth 2 again [s5, a0, b76] Result: 5
3 5 0 2397 e2e4 d7d5 f1b5 c8d7 b5d3 d5e4 d3e4
3 5 0 2397 e2e4 d7d5 f1b5 c8d7 b5d3 d5e4 d3e4
4 30 1 5705 e2e4 e7e6 d2d4 d7d5 f1d3 d5e4 d3e4
4 30 1 7516 e2e4 e7e6 d2d4 d7d5 f1d3 d5e4 d3e4
5 7 3 18893 e2e4 d7d5 e4e5 d5d4 f1b5 c8d7 b5d3
5 7 4 21369 e2e4 d7d5 e4e5 d5d4 f1b5 c8d7 b5d3
6 11 7 39287 e2e4 d7d5 e4e5 d5d4 f1c4 e7e6 d2d3
6 11 10 64113 e2e4 d7d5 e4e5 d5d4 f1c4 e7e6 d2d3
7 12 32 185152 e2e4 d7d5 e4d5 d8d5 b1c3 d5d4 f1b5 c7c6 g1f3
7 12 35 204390 e2e4 d7d5 e4d5 d8d5 b1c3 d5d4 f1b5 c7c6 g1f3
8 7 53 310170 e2e4 d7d5 e4d5 d8d5 b1c3 d5e5 d1e2 b8c6 g1f3 e5d6
Searching depth 7 again [s8, a7, b63] Result: 13
8 13 84 490395 d2d4 d7d5 e2e3 e7e6 b1c3 b8c6 g1f3 g8f6 f1d3
8 13 84 493709 d2d4 d7d5 e2e3 e7e6 b1c3 b8c6 g1f3 g8f6 f1d3
9 5 167 977505 d2d4 d7d5 e2e3 e7e6 g1f3 g8f6 f1b5 c7c6 b5d3 f8b4 c1d2 b4d6
Searching depth 8 again [s6, a5, b64] Result: 9
9 9 289 1691747 e2e4 d7d5 e4d5 d8d5 b1c3 d5d6 d2d4 e7e5 d4e5 d6e5 c1e3
9 9 306 1790835 e2e4 d7d5 e4d5 d8d5 b1c3 d5d6 d2d4 e7e5 d4e5 d6e5 c1e3
10 14 476 2800071 e2e4 d7d5 e4d5 d8d5 b1c3 d5e5 f1e2 b8c6 g1f3 e5a5 d2d4 e7e5
10 14 571 3366085 e2e4 d7d5 e4d5 d8d5 b1c3 d5e5 f1e2 b8c6 g1f3 e5a5 d2d4 e7e5
11 13 896 5266788 e2e4 d7d5 e4d5 d8d5 b1c3 d5a5 d2d4 e7e5 d1e2 f8d6 f2f4 f7f6 f4e5
11 13 1415 8378552 e2e4 d7d5 e4d5 d8d5 b1c3 d5a5 d2d4 e7e5 d1e2 f8d6 f2f4 f7f6 f4e5
12 21 4106 24109863 e2e4 e7e5 b1c3 g8f6 g1f3 f8d6 d2d4 e5d4 f3d4 e8g8 f1d3 b8c6 c1e3 c6d4
12 21 4361 25632458 e2e4 e7e5 b1c3 g8f6 g1f3 f8d6 d2d4 e5d4 f3d4 e8g8 f1d3 b8c6 c1e3 c6d4



my ftp site {remove http:// unless you like error messages}

[Dann Corbit]

Geschrieben von:/Posted by: Dann Corbit at 24 September 2004 07:31:54:
Als Antwort auf:/In reply to: With big hash Natwarlal v0.06 searches pretty fast... geschrieben von:/posted by: Dann Corbit at 24 September 2004 07:03:17:

Natwarlal v0.06
by Pallav Nawani, India.
Natwarlal is a winboard compatible chess engine. Natwarlal is released under
MIT license. See license.txt (included with this program) for details.
Type 'help' for a list of commands.
Initialized Hashtables.
Hashtable 131072K, Pawn hashtable 32768K
Opening book initialized (from natwarlal.bk).
Positions in book: 43579
Ready to roll! (PS: Type 'new' to start a game)
Natwarlal> st 999
Natwarlal> sd 999
Natwarlal> post
Natwarlal> new
Black
8 r n b q k b n r
7 p p p p p p p p
6 - - - - - - - -
5 - - - - - - - -
4 - - - - - - - -
3 - - - - - - - -
2 P P P P P P P P
1 R N B Q K B N R
a b c d e f g h
White
White to move [KQkq[-]
Natwarlal> go
Time alotted 3600s, ext 1200
1 0 0 21 d2d4 d7d5
1 0 0 65 d2d4 d7d5
2 25 0 141 d2d4 d7d5 e2e3
2 25 0 651 d2d4 d7d5 e2e3
3 0 0 807 d2d4 d7d5 e2e3 e7e6
Searching depth 2 again [s5, a0, b76] Result: 5
3 5 0 2397 e2e4 d7d5 f1b5 c8d7 b5d3 d5e4 d3e4
3 5 0 2397 e2e4 d7d5 f1b5 c8d7 b5d3 d5e4 d3e4
4 30 1 5705 e2e4 e7e6 d2d4 d7d5 f1d3 d5e4 d3e4
4 30 1 7516 e2e4 e7e6 d2d4 d7d5 f1d3 d5e4 d3e4
5 7 3 18893 e2e4 d7d5 e4e5 d5d4 f1b5 c8d7 b5d3
5 7 4 21369 e2e4 d7d5 e4e5 d5d4 f1b5 c8d7 b5d3
6 11 7 39287 e2e4 d7d5 e4e5 d5d4 f1c4 e7e6 d2d3
6 11 10 64113 e2e4 d7d5 e4e5 d5d4 f1c4 e7e6 d2d3
7 12 32 185152 e2e4 d7d5 e4d5 d8d5 b1c3 d5d4 f1b5 c7c6 g1f3
7 12 35 204390 e2e4 d7d5 e4d5 d8d5 b1c3 d5d4 f1b5 c7c6 g1f3
8 7 53 310170 e2e4 d7d5 e4d5 d8d5 b1c3 d5e5 d1e2 b8c6 g1f3 e5d6
Searching depth 7 again [s8, a7, b63] Result: 13
8 13 84 490395 d2d4 d7d5 e2e3 e7e6 b1c3 b8c6 g1f3 g8f6 f1d3
8 13 84 493709 d2d4 d7d5 e2e3 e7e6 b1c3 b8c6 g1f3 g8f6 f1d3
9 5 167 977505 d2d4 d7d5 e2e3 e7e6 g1f3 g8f6 f1b5 c7c6 b5d3 f8b4 c1d2 b4d6
Searching depth 8 again [s6, a5, b64] Result: 9
9 9 289 1691747 e2e4 d7d5 e4d5 d8d5 b1c3 d5d6 d2d4 e7e5 d4e5 d6e5 c1e3
9 9 306 1790835 e2e4 d7d5 e4d5 d8d5 b1c3 d5d6 d2d4 e7e5 d4e5 d6e5 c1e3
10 14 476 2800071 e2e4 d7d5 e4d5 d8d5 b1c3 d5e5 f1e2 b8c6 g1f3 e5a5 d2d4 e7e5
10 14 571 3366085 e2e4 d7d5 e4d5 d8d5 b1c3 d5e5 f1e2 b8c6 g1f3 e5a5 d2d4 e7e5
11 13 896 5266788 e2e4 d7d5 e4d5 d8d5 b1c3 d5a5 d2d4 e7e5 d1e2 f8d6 f2f4 f7f6 f4e5
11 13 1415 8378552 e2e4 d7d5 e4d5 d8d5 b1c3 d5a5 d2d4 e7e5 d1e2 f8d6 f2f4 f7f6 f4e5
12 21 4106 24109863 e2e4 e7e5 b1c3 g8f6 g1f3 f8d6 d2d4 e5d4 f3d4 e8g8 f1d3 b8c6 c1e3 c6d4
12 21 4361 25632458 e2e4 e7e5 b1c3 g8f6 g1f3 f8d6 d2d4 e5d4 f3d4 e8g8 f1d3 b8c6 c1e3 c6d4

I sometimes do a game I call "what would he be thinking..." that shows the name of the game the program is considering at the root position. (Crafty used to choose the dragon long ago, but I have not tried recently). Here is Natwarlal's notion:
[Event "Edited game"]
[Site "DCORBIT64"]
[Date "2004.09.23"]
[Round "-"]
[White "-"]
[Black "-"]
[Result "*"]
[ECO "C42c"]
[Variation "Russian-Three Knights Game"]
1. e4 e5 2. Nc3 Nf6 3. Nf3 Bd6 4. d4 exd4 5. Nxd4 O-O 6. Bd3 Nc6 7. Be3 Nxd4 *
This is Crafty 19.17 SE (Fischer) taking a look at the EPD:
rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - acd 13; acn 50533643; acs 60; bm e4; ce 7; pv Nf3 Nf6 e3 d5 Be2 Nc6 O-O e5 d4 e4 Ne5 Bd6 Bb5;
rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - acd 13; acn 47961153; acs 60; bm e5; ce -21; pv e5 Nf3 Nf6 Bc4 Nc6 d4 Bd6 O-O O-O d5 Na5 Bd3;
rnbqkbnr/pppp1ppp/8/4p3/4P3/8/PPPP1PPP/RNBQKBNR w KQkq - acd 13; acn 51243954; acs 60; bm Nc3; ce 7; pv Nf3 Nf6 Nxe5 Qe7 Nf3 Qxe4+ Be2 Bc5 Nc3 Qg4 O-O O-O Ne5;
rnbqkbnr/pppp1ppp/8/4p3/4P3/2N5/PPPP1PPP/R1BQKBNR b KQkq - acd 13; acn 50785616; acs 60; bm Nf6; ce -13; pv Nf6 Bc4 Nc6 Nf3 Bc5 O-O d6 d3 O-O Bg5 Bg4 Nd5;
rnbqkb1r/pppp1ppp/5n2/4p3/4P3/2N5/PPPP1PPP/R1BQKBNR w KQkq - acd 12; acn 47938067; acs 60; bm Nf3; ce -6; pv Nf3 Bb4 Nxe5 O-O Be2 Bxc3 dxc3 Re8 Nd3 Nxe4 O-O Qf6 Bf4;
rnbqkb1r/pppp1ppp/5n2/4p3/4P3/2N2N2/PPPP1PPP/R1BQKB1R b KQkq - acd 13; acn 50391420; acs 61; bm Bd6; ce -14; pv Bb4 a3 Bxc3 dxc3 O-O Bg5 d6 Qe2 Na6 O-O-O Bd7 Qe3 Ng4 Qe2;
rnbqk2r/pppp1ppp/3b1n2/4p3/4P3/2N2N2/PPPP1PPP/R1BQKB1R w KQkq - acd 14; acn 51281737; acs 60; bm d4; ce 5; pv Bc4 O-O O-O Nc6 Re1 Bc5 d3 d6 Bg5 Bb4 d4 Bxc3 bxc3 Bg4 dxe5 Nxe5;
rnbqk2r/pppp1ppp/3b1n2/4p3/3PP3/2N2N2/PPP2PPP/R1BQKB1R b KQkq - acd 13; acn 56745563; acs 61; bm exd4; ce -29; pv exd4 Qxd4 Nc6 Qc4 O-O Bg5 Bb4 e5 Bxc3+ Qxc3 Qe8 O-O-O Ng4 Bc4 Ncxe5;
rnbqk2r/pppp1ppp/3b1n2/8/3pP3/2N2N2/PPP2PPP/R1BQKB1R w KQkq - acd 13; acn 56532378; acs 60; bm Nxd4; ce -19; pv Qxd4 O-O Bc4 Nc6 Qd3 Ne5 Nxe5 Bxe5 O-O Qe7 Nd5 Nxd5 Qxd5 d6;
rnbqk2r/pppp1ppp/3b1n2/8/3NP3/2N5/PPP2PPP/R1BQKB1R b KQkq - acd 13; acn 52818145; acs 60; bm O-O; ce 2; pv O-O Bg5 Nc6 Nf5 Be5 Qf3 d5 O-O-O Bxc3 Qxc3 Bxf5 Bxf6 Qxf6 Qxf6 gxf6 exf5;
rnbq1rk1/pppp1ppp/3b1n2/8/3NP3/2N5/PPP2PPP/R1BQKB1R w KQ - acd 12; acn 54973600; acs 61; bm Bd3; ce -17; pv Bd3 Be5 Nde2 d5 exd5 Ng4 h3 Nh2 Bf4 Bxf4 Nxf4 Re8+ Nce2 g5;
rnbq1rk1/pppp1ppp/3b1n2/8/3NP3/2NB4/PPP2PPP/R1BQK2R b KQ - acd 12; acn 54433135; acs 62; bm Nc6; ce -3; pv Nc6 Ndb5 Bb4 O-O a6 Na3 Qe7 Nd5 Nxd5 exd5 Qe5 dxc6 Bxa3;
Even these last two (Qsearch) look pretty much OK:
r1bq1rk1/pppp1ppp/2nb1n2/8/3NP3/2NB4/PPP2PPP/R1BQK2R w KQ - acd 13; acn 54857366; acs 61; bm Be3; ce 11; pv Nf3 Bb4 O-O Re8 Re1 d6 Bf4 Bg4 h3 Bxf3 Qxf3 Nd4 Qg3 Bxc3 bxc3;
r1bq1rk1/pppp1ppp/2nb1n2/8/3NP3/2NBB3/PPP2PPP/R2QK2R b KQ - acd 12; acn 55340805; acs 61; bm Nxd4; ce -2; pv Be5 Nxc6 bxc6 O-O Re8 Re1 Bxc3 bxc3 Nxe4 Bf4 Nxc3 Bxh7+ Kxh7 Qd3+ Kg8 Qxc3;



my ftp site {remove http:// unless you like error messages}

[Pallav Nawani]

Geschrieben von:/Posted by: Pallav Nawani at 24 September 2004 20:25:32:
Als Antwort auf:/In reply to: Re: With big hash Natwarlal v0.06 searches pretty fast... geschrieben von:/posted by: Dann Corbit at 24 September 2004 07:31:54:

Natwarlal v0.06

I sometimes do a game I call "what would he be thinking..." that shows the name of the game the program is considering at the root position. (Crafty used to choose the dragon long ago, but I have not tried recently). Here is Natwarlal's notion:
[Event "Edited game"]
[Site "DCORBIT64"]
[Date "2004.09.23"]
[Round "-"]
[White "-"]
[Black "-"]
[Result "*"]
[ECO "C42c"]
[Variation "Russian-Three Knights Game"]
1. e4 e5 2. Nc3 Nf6 3. Nf3 Bd6 4. d4 exd4 5. Nxd4 O-O 6. Bd3 Nc6 7. Be3 Nxd4 *
This is Crafty 19.17 SE (Fischer) taking a look at the EPD:

Hi,
That was quick! How'd you know a new version was released? Some automatic website monitoring system?
I couldn't quite get what's going on here. You analyzed the root position with Natwarlal, and it looks like the line it took is called Russian-Three Knights Game. After that I couldn't follow you. What were those EPD's and what was the qsearch EPD about?

Thanks,
Pallav

[Dann Corbit]

Geschrieben von:/Posted by: Dann Corbit at 24 September 2004 20:50:19:
Als Antwort auf:/In reply to: Re: With big hash Natwarlal v0.06 searches pretty fast... geschrieben von:/posted by: Pallav Nawani at 24 September 2004 20:25:32:

Natwarlal v0.06

I sometimes do a game I call "what would he be thinking..." that shows the name of the game the program is considering at the root position. (Crafty used to choose the dragon long ago, but I have not tried recently). Here is Natwarlal's notion:
[Event "Edited game"]
[Site "DCORBIT64"]
[Date "2004.09.23"]
[Round "-"]
[White "-"]
[Black "-"]
[Result "*"]
[ECO "C42c"]
[Variation "Russian-Three Knights Game"]
1. e4 e5 2. Nc3 Nf6 3. Nf3 Bd6 4. d4 exd4 5. Nxd4 O-O 6. Bd3 Nc6 7. Be3 Nxd4 *
This is Crafty 19.17 SE (Fischer) taking a look at the EPD:

Hi,
That was quick! How'd you know a new version was released? Some automatic website monitoring system?
I couldn't quite get what's going on here. You analyzed the root position with Natwarlal, and it looks like the line it took is called Russian-Three Knights Game. After that I couldn't follow you.
What were those EPD's and what was the qsearch EPD about?

I think I saw it on Leo's page, but maybe I web search of some kind. I forget.
The EPD positions are the moves your program was considering at the root. I analyzed them with a version of crafty, and every move in the PV is playable. That is an important piece of debugging information. It means that your PV is reliable and that is a very good thing.
It was a 12 ply search. Towards the end of 12 plies, things often get more and more iffy. And for the 13th half-move onward, the nodes displayed are not found by regular search but by q-search, since the search performed was only 12 plies from the root. Because the PV was still sensible past the last searched node, I think you can be happy that your search is sensible, including the qsearch.
Sometimes you will find engines do wild things, especially towards the end of the search and very often you will see wild insanity in the qsearched nodes.



my ftp site {remove http:// unless you like error messages}

[Pallav Nawani]

Geschrieben von:/Posted by: Pallav Nawani at 24 September 2004 21:10:09:
Als Antwort auf:/In reply to: Re: With big hash Natwarlal v0.06 searches pretty fast... geschrieben von:/posted by: Dann Corbit at 24 September 2004 20:50:19:

Natwarlal v0.06

I sometimes do a game I call "what would he be thinking..." that shows >>>[Variation "Russian-Three Knights Game"]
1. e4 e5 2. Nc3 Nf6 3. Nf3 Bd6 4. d4 exd4 5. Nxd4 O-O 6. Bd3 Nc6 7. Be3 Nxd4 *
This is Crafty 19.17 SE (Fischer) taking a look at the EPD:

Hi,
That was quick! How'd you know a new version was released? Some automatic website monitoring system?
I couldn't quite get what's going on here. You analyzed the root position with Natwarlal, and it looks like the line it took is called Russian-Three Knights Game. After that I couldn't follow you.
What were those EPD's and what was the qsearch EPD about?

I think I saw it on Leo's page, but maybe I web search of some kind. I forget.
The EPD positions are the moves your program was considering at the root. I analyzed them with a version of crafty, and every move in the PV is playable. That is an important piece of debugging information. It means that your PV is reliable and that is a very good thing.
It was a 12 ply search. Towards the end of 12 plies, things often get more and more iffy. And for the 13th half-move onward, the nodes displayed are not found by regular search but by q-search, since the search performed was only 12 plies from the root. Because the PV was still sensible past the last searched node, I think you can be happy that your search is sensible, including the qsearch.
Sometimes you will find engines do wild things, especially towards the end of the search and very often you will see wild insanity in the qsearched nodes.

I understand now. That's a very useful way of analyzing PVs.
Best regards,
Pallav

[Dann Corbit]

Geschrieben von:/Posted by: Dann Corbit at 24 September 2004 21:48:37:
Als Antwort auf:/In reply to: Re: With big hash Natwarlal v0.06 searches pretty fast... geschrieben von:/posted by: Pallav Nawani at 24 September 2004 21:10:09:

Natwarlal v0.06

I sometimes do a game I call "what would he be thinking..." that shows >>>[Variation "Russian-Three Knights Game"]
1. e4 e5 2. Nc3 Nf6 3. Nf3 Bd6 4. d4 exd4 5. Nxd4 O-O 6. Bd3 Nc6 7. Be3 Nxd4 *
This is Crafty 19.17 SE (Fischer) taking a look at the EPD:

Hi,
That was quick! How'd you know a new version was released? Some automatic website monitoring system?
I couldn't quite get what's going on here. You analyzed the root position with Natwarlal, and it looks like the line it took is called Russian-Three Knights Game. After that I couldn't follow you.
What were those EPD's and what was the qsearch EPD about?

I think I saw it on Leo's page, but maybe I web search of some kind. I forget.
The EPD positions are the moves your program was considering at the root. I analyzed them with a version of crafty, and every move in the PV is playable. That is an important piece of debugging information. It means that your PV is reliable and that is a very good thing.
It was a 12 ply search. Towards the end of 12 plies, things often get more and more iffy. And for the 13th half-move onward, the nodes displayed are not found by regular search but by q-search, since the search performed was only 12 plies from the root. Because the PV was still sensible past the last searched node, I think you can be happy that your search is sensible, including the qsearch.
Sometimes you will find engines do wild things, especially towards the end of the search and very often you will see wild insanity in the qsearched nodes.

I understand now. That's a very useful way of analyzing PVs.

I think somehow it also tells us a chess engine's personality (or perhaps the author's). We see if the engine might try a gambit or something loose like that. Will the engine choose the d or e pawn, or one of the knights, or something wild and crazy? I am still looking for the engine that likes 1. b4




my ftp site {remove http:// unless you like error messages}