Lime_v48

[Richard Allbert]

Hi,

Those of you who like weak engines, Lime_v48 has now been released, changes below.

Code: Select all

v48
=====================================================================================
-   *added insufficient material draw claim in xboard mode
-   *added mate claim in xboard mode
-   *fixed promotion bug in xboard mode (underpromoted piece not printed
   e.g c2c1 now reads c2c1n)
-   *fixed a bad en-passant bug - during takeback, if the move was not an
   ep-capture, the engine changed the ep-square to null
-   *fixed bug in hash tables when returning score
-   *fixed ep bug in fen parsing (used '=' instead of '==')
-   *fixed book use in UCI mode
-   *fixed hash changes in null move - not hashing ep sqaure or side correctly
-   compiled with g++.exe -O and had a huge increase in nps. Won't compile with -O2 (yet)
-   removed epd test function
-   added perft testing, passes all tests
-   added benchmark function
-   added Lime.ini for specifying hash size
-   added bool type to hash structure to detect null move permission
-   removed 'force e4' as first move
-   added makebook function, now uses its own original book
-   completely revised book code - reads the bookfile into an internal STL vector when the
   program boots (occupies 4MB with default book)
-   added pawn on 7th rank extension
-   now updates hash during makemove; previous versions scanned the whole board
-   added recognition of 'result' in xboard mode to stop draw claims
   continuing past the end of the game
-   changed move generation for castling - cascaded the if()'s
-   added mirror function for eval testing, fixed bugs in engame eval() scoring
-   made some changes to evaluation
-   now using 64 bit hash keys
-   added time handling for 'moves per session' time controls
-   changed piece values - no longer standard 100, 300, 300, 500,900


The changes were mainly to iron out the bugs. This will be the last release for some time, as I am part way through a complete rewrite of a completely new (and hopefully more sophisticated) version.

Just one note for users that don't read the readme.... Lime reads the book into a vector on startup, which occupies about 4MB RAM.

Regards,

Richard

[Roger Brown]

Hello Richard,

I like weak engines.

Thanks.

However, I am a cynic.

What does weak mean?



Other authors mean weak in a less than 2200 context.

Weak for me means about 900 elo.



Later.

[Richard Allbert]

Hi Roger

Weak as in 1800 elo (on a good day)

v21 on my site is about 1500...

Regards,

Richard
www.limechess.homecall.co.uk

[Patrick Buchmann]

Hi,

Lime v48 obtains 1982 as first rating in the UEL

Code: Select all

Tournoi d'entr?e BOHScc.abk
Le Fou numerique, 2005.11.12 - 2005.11.13
                          Score       1    2
----------------------------------------------
 1: Lime v48             6.0 / 84   35w- 35b-
 2: Abrok 5.0            2.0 /  2    1b+  1w+
 3: Aice 0.97            2.0 /  2    1b+  1w+
 4: Amyan 1.597          2.0 /  2    1b+  1w+
 5: AnMon 5.53           2.0 /  2    1b+  1w+
 6: Arion 1.6            2.0 /  2    1b+  1w+
 7: Aristarch 4.50       2.0 /  2    1b+  1w+
 8: Booot 4.75           2.0 /  2    1b+  1w+
 9: Delphil 1.5b         2.0 /  2    1b+  1w+
10: Diablo 0.3           2.0 /  2    1b+  1w+
11: Dragon 4.6           2.0 /  2    1b+  1w+
12: Frenzee 200          2.0 /  2    1b+  1w+
13: Fruit 2.1            2.0 /  2    1b+  1w+
14: Glaurung Mainz       2.0 /  2    1b+  1w+
15: Hermann 1.5          2.0 /  2    1b+  1w+
16: Homer 1.02           2.0 /  2    1b+  1w+
17: List 5.12            2.0 /  2    1b+  1w+
18: Madeleine 0.2        2.0 /  2    1b+  1w+
19: Naum 1.8             2.0 /  2    1b+  1w+
20: Nejmet 3.07          2.0 /  2    1b+  1w+
21: Patzer 3.61          2.0 /  2    1b+  1w+
22: Pepito v1.59         2.0 /  2    1b+  1w+
23: Pharaon 3.3          2.0 /  2    1b+  1w+
24: Queen 3.07           2.0 /  2    1b+  1w+
25: Ruffian 1.0.5        2.0 /  2    1b+  1w+
26: SlowChess Blitz WV   2.0 /  2    1b+  1w+
27: Snitch 1.0.8         2.0 /  2    1b+  1w+
28: SOS 5.1              2.0 /  2    1b+  1w+
29: Spike 1.0a Mainz     2.0 /  2    1b+  1w+
30: Tao 5.6              2.0 /  2    1b+  1w+
31: Terra 3.4            2.0 /  2    1b+  1w+
32: Tornado 0.84.2       2.0 /  2    1b+  1w+
33: Ufim 7.01            2.0 /  2    1b+  1w+
34: Yace Paderborn       2.0 /  2    1b+  1w+
35: Zappa 1.1            2.0 /  2    1b+  1w+
36: Muse 0.899b          1.5 /  2    1b+  1w=
37: WJChess v1.64        1.5 /  2    1b+  1w=
38: Ayito 0.2.994        1.5 /  2    1b=  1w+
39: Chispa 4.0.3         1.5 /  2    1b=  1w+
40: Sjeng 12.13          1.0 /  2    1b+  1w-
41: Adam 2.9             1.0 /  2    1b-  1w+
42: Ktulu 4.2            1.0 /  2    1b+  1w-
43: Gaia 2.2a            1.0 /  2    1b-  1w+
----------------------------------------------
84 games: +40 =4 -40

All games with the book BOHScc

PGN on my homepage

Regards,
Patrick

[Richard Allbert]

Thanks, Patrick.

I expected a little more than that ; should eventually settle 60-80 points higher than v41.

Regards,

Richard

[Patrick Buchmann]

Thanks, Patrick.

I expected a little more than that ; should eventually settle 60-80 points higher than v41.

Regards,

Richard

Hi Richard,

v41 made a tournament with other "weak" engines and was on the 10th place (22 engines).

v48 had just play 84 games here.

Regards,
Patrick[/url]

[Richard Allbert]

Ok,

Thanks again!

Regards,

Richard