Tucano chess engine 1.00

[jshriver]

So far it seems very nice, I've enjoyed watching the games.

-Josh

[Graham Banks]

Hello, I made available my chess engine with source code. For now I just have the windows 32 bit executable. I implemented xboard protocol.

I've been working for some years in my spare time, and I choose the name tucano (or toucan in english) because I want something related to my country Brazil.

Thanks to everyone that contribute to chess programming, mainly CPW, talkchess, and all available source codes.

I implemented a basic bitboard with no magic, rotated style. I left some of my testing routines that my help others, such as epd test, eval test and 3 types of perft.

Hope you enjoy. This is a very addictive hobby. I'm planning to keep working on it.

Thanks
Alcides.

https://sites.google.com/site/tucanochess/

Roughly how strong do you estimate that your engine is?

[jshriver]

Roughly how strong do you estimate that your engine is?

It tied with Fruit in a recent bullet game. Haven't tried with long timecontrol yet though.

tucanojs (1626) vs. fruitjs (1919) ---
Rated lightning match, initial time: 1 minutes, increment: 0 seconds.

Move tucanojs fruitjs
---- ---------------- ----------------
1. d4 (0:00) Nf6 (0:00)
2. c4 (0:00) e6 (0:00)
3. Nc3 (0:00) Bb4 (0:00)
4. Qc2 (0:00) d5 (0:00)
5. cxd5 (0:00) Qxd5 (0:00)
6. e3 (0:00) c5 (0:00)
7. Bd2 (0:00) Bxc3 (0:00)
8. Bxc3 (0:00) cxd4 (0:00)
9. Bxd4 (0:00) Nc6 (0:00)
10. Bc3 (0:00) O-O (0:00)
11. Nf3 (0:02) Rd8 (0:00)
12. Be2 (0:02) Qe4 (0:00)
13. Qb3 (0:02) Nd5 (0:00)
14. O-O (0:02) Nxc3 (0:02)
15. Qxc3 (0:01) e5 (0:01)
16. Bc4 (0:01) Be6 (0:00)
17. Bxe6 (0:01) fxe6 (0:00)
18. Qb3 (0:02) Qd5 (0:00)
19. Qa4 (0:01) b5 (0:01)
20. Qg4 (0:02) h6 (0:00)
21. Nh4 (0:02) Rac8 (0:01)
22. Rfd1 (0:02) Qc4 (0:03)
23. Qxc4 (0:01) bxc4 (0:00)
24. Kf1 (0:02) e4 (0:03)
25. Ng6 (0:02) Kf7 (0:01)
26. Nf4 (0:01) Ne5 (0:01)
27. Rxd8 (0:02) Rxd8 (0:00)
28. Ne2 (0:01) Rb8 (0:02)
29. b3 (0:01) cxb3 (0:02)
30. axb3 (0:01) Rxb3 (0:01)
31. Rxa7+ (0:01) Kg6 (0:00)
32. Nf4+ (0:01) Kf6 (0:02)
33. Nh5+ (0:01) Kg6 (0:02)
34. Nxg7 (0:01) Rb1+ (0:01)
35. Ke2 (0:01) Rb2+ (0:01)
36. Kf1 (0:01) Rb1+ (0:01)
37. Ke2 (0:01) Rb2+ (0:01)
38. Kf1 (0:01) Rb1+ (0:05)
{Game drawn by repetition} 1/2-1/2

[sedicla]

Roughly how strong do you estimate that your engine is?

Hi, my guesstimate is around 2400-2500 ???

Here is the final standings for the last 1m1s gaunlet. I usually play against those engines, and get 50-55% points.

By the way, if possible i'ld like to see my engine in one of your tournaments. Let me know if you need anything. Thanks.



Engine Score Tu

Code: Select all

1: Tucano             27.5/50 ·········· 
2: Rotor              6.5/10  11111==0=0 
3: Ruffian 1.0.5      6.0/10  01=1==01=1 
4: Rodent             4.5/10  11=1010000 
4: Hermann 2.6 32 bit 4.5/10  =1100===0= 
6: Olithink           1.0/10  0001000000 

50 games played / Tournament is finished
Name of the tournament: T004
Site/ Country: TOSHIBA-LAPTOP, United States
Level: Blitz 1/1

[sedicla]

Roughly how strong do you estimate that your engine is?

It tied with Fruit in a recent bullet game. Haven't tried with long timecontrol yet though.

Sometimes I play a 100 games 10s1ms against fruit and just make a scratch, like about 10% win. Fruit is so well coded that it hurts , and it is in another league.
Thanks.

[Graham Banks]

Roughly how strong do you estimate that your engine is?

Hi, my guesstimate is around 2400-2500 ???

Here is the final standings for the last 1m1s gaunlet. I usually play against those engines, and get 50-55% points.

By the way, if possible i'ld like to see my engine in one of your tournaments. Let me know if you need anything. Thanks.



Engine Score Tu

Code: Select all

1: Tucano             27.5/50 ·········· 
2: Rotor              6.5/10  11111==0=0 
3: Ruffian 1.0.5      6.0/10  01=1==01=1 
4: Rodent             4.5/10  11=1010000 
4: Hermann 2.6 32 bit 4.5/10  =1100===0= 
6: Olithink           1.0/10  0001000000 

50 games played / Tournament is finished
Name of the tournament: T004
Site/ Country: TOSHIBA-LAPTOP, United States
Level: Blitz 1/1

That gives me a rough idea. Thanks.

My intention was to include it in my Amateur Series tournaments.
It looks like the upcoming Division 5 might be suitable. Engines in that field are likely to be Rodin, Delphil, Murka, Cheese, iCE, Dorky, Eeyore, Ifrit, Betsabe, Hussar and OliThink.

[sedicla]

Yes you can post there, thanks.
I wonder what went wrong, if you send me the changes you made I can take a look and include in my original.

Most notoriously, don't use plain "char" if you need it to be either signed or unsigned. Simply make the full type explicit.
In general, I would strongly recommend you compile and run/test your code on Linux to deal with a few other portability issues.
I am happy to test a new version once you have one.

Hi I uploaded a new version with some compatibility fixes and some other minor changes. Below is the release notes. I also included an executable for Linux (Ubuntu 12.04.LTS). I don't know if it is compatible with other linux versions. But should compile now for other systems.

Thanks.



1.04
* Compatibility fixes: review of some variables declaration in order to make possible to compile for linux and hopefuly other systems (ARM, MAC). Changes made mainly at bitboard.c and globals.h.
* Added a preprocessor test to not use BBIX structure when system has different size for unsigned int and unsigned short data types at bitboard.c.
* Implemented hash table size parameter -hash. Also changed the default to be 64MB instead of 128MB which was the initial idea.
* Made a small optimization to keep king square as integer in the board.state structure. Only recalculate when king moves.
* Fixed the perft time calculation, was using clock() method. Changed to use the same functions as the search.
* Added the parameter "-nobook" to turn off the small internal opening book.
* Included Linux executable for Ubuntu 12.04 LTS.
* Added a .jpg image for program icon.

1.00
* initial version
* windows 32 executable

[abik]

Hi I uploaded a new version with some compatibility fixes and some other minor changes.

Thanks. Per your prior permission, I put an ARM binary of tucano104 that runs smoothly in Chess for Android on UCI and XBoard engines for Android.
A quick 1s/move test run with random opening book against BikJump ended in 7-3 in favor of Tucano!