Toga II 3.1.2SE

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
Graham Banks
Posts: 45587
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Toga II 3.1.2SE

Post by Graham Banks »

Toga II 3.1.2SE including source

Following the release of the source the SE versions are now included in the 40/4 rating list

Single-CPU engines

About Toga II 3.1.2SE

Modifications to Toga by Chris Formula

Readme

LEGAL DETAILS
-------------------------

This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or (at
your option) any later version.

This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
USA

See the file "copying.txt" for details.


GENERAL
----------------

The Special Edition or "SE" of Toga II based on Fruit 2.1 is being developed to
further increase the playing strength of the engine created by Thomas Gaksch.
Though Toga II is among the topnotch chess engines (and FREE) today, there
are still rooms for improvement on its search and evaluation. When Thomas
showed signs to quit computer chess, the "SE" project came to light to
continue what has been started by him. As of the moment, the "SE" project is
being done privately by Chris Formula.


EPD TESTING
----------------------

EET testsuite of Walter Eigenmann was used to evaluate improvement in the endgame
knowledge. Such testsuite was composed of 100 important test positions. Moreover, the
positional prowess of the engine was evaluated using the SSDF Positional testsuite of
Tony Hedlund. Such testsuite was composed of 16 important test positions. The total
points and the number of best move found (in parenthesis) were recorded. Below are
settings used and the results obtained during EPD testing. Amazing results!

Hash: 128M
Time Limit: 60 sec/move for EET; 600 sec/move for SSDF Positional
Software: Arena by Martin Blume
Bitbases: 3-4-man Scorpiobases by Daniel Shawul
Success Criteria:
(1) at least 50 points in EET testsuite
(2) at least 128 points in SSDF Positional testsuite

Version EET Testsuite SSDF Positional
1.3.1 48/100 116/160 (8/16)
1.3.4 45/100 105/160 (7/16)
2.1SE 51/100 117/160 (8/16)
2.2SE 53/100 88/160 (4/16)
3.1.2SE* 50/100* 128/160 (8/16)*


VERSION HISTORY
------------------------------

Toga II 3.1.2SE based on Fruit 2.1
- code was based on Toga II 1.3.1 based on Fruit 2.1
- improved search efficiency
- play style becomes more aggressive

Toga II 2.2SE based on Fruit 2.1
- code was based on Toga II 1.3.4 based on Fruit 2.1
- improved material evaluation

Toga II 2.1SE based on Fruit 2.1
- code was based on Toga II 1.3.1 based on Fruit 2.1
- improved material evaluation

Toga II 1.3.4 based on Fruit 2.1
Following changes and extensions made:
- improved history pruning
- improved futility pruning
- improved lazy evaluation
- new evaluation features (outpost knight etc.)
- MultiPV Mode
- This version performs similar level as version 1.3.1. However, its endgame is more accurate and uses EGBBs provided by Daniel Shawul, author of Scorpio.
- This version was previously known as 1.3x4b with some setting tweaks as follows.
a) Toga Extended History Pruning has been turned off
b) History Threshold has been adjusted to 70
- For Blitz play Toga Extended History Pruning should be turned on and the History Threshold adjusted to 80. These setting have shown to improve performance at blitz.

Toga II 1.3x4b based on Fruit 2.1
- bugfix of the endgame bitbases with help from Daniel Shawul

Toga II 1.3x4 based on Fruit 2.1
- improved evaluation
- improved search
- added support to endgame bitbases by Daniel Shawul

Toga II 1.3.1 based on Fruit 2.1
Following changes and extensions made:
- improved history pruning
- improved futility pruning
- improved lazy evaluation
- new evaluation features (outpost knight etc.)
- MultiPV Mode
- This is the best performing beta provided by Thomas and release with his permission.
- This version was previously known as 1.3beta1.
- For Blitz play Toga Extended History Pruning should be turned on and the History Threshold adjusted to 75. These settings have shown to improve performance at blitz.

Toga II 1.2.1a based on Fruit 2.1
- problems with internal opening book fixed
- playing strength is 100% identically to 1.2.1

Toga II 1.2.1 based on Fruit 2.1
- new settings
- some minor bugfixes

Toga II 1.2 based on Fruit 2.1
- more pruning
- more extensions
- new settings by Dieter Eberle


EGBBs Installation
----------------------------
Toga II uses the Scorpio endgame bitbases up to 5 pieces.

1. Download the 5-man bitbases from Leo Dijksman's WBEC site http://www.wbec-ridderkerk.nl . The egbbs are 340mb in size.
2. Store them anywhere in your computer. The default path is c:\egbb\ but you can change this in the uci options.
IMPORTANT: The egbbdll.dll must be in the same folder as the bitbase files.


ACKNOWLEDGMENT
--------------------------------

Huge thanks to Fabien Letouzey (France) for the great source code of the program Fruit 2.1.
Huge thanks to Thomas Gaksch (Germany) for the great source code of the program Toga II based on Fruit 2.1.

Special thanks to:
Dieter Eberle, Karl-Heinz Söntges, Shaun Brewer for testing Beta Versions.
A big thanks to Dieter Eberle for his setting.
Orlando Mouchel for bugfixes and new ideas.
Wilhelm Hudetz for the Logo.
Daniel Shawul for sharing his great endgame bitbases and his dll.
Walter Eigenmann for the EET testsuite.
Tony Hedlund for the SSDF Positional testsuite.
Martin Blume for the Arena software.
Denis Mendoza for the compile support for different platforms.

Without their help, the improved playing strength of Toga II based on Fruit 2.1 would not be possible.
Thank you! Maraming salamat!

Regards,
Thomas Gaksch (Germany)
Chris Formula (Philippines)
gbanksnz at gmail.com
User avatar
Mike S.
Posts: 1480
Joined: Thu Mar 09, 2006 5:33 am

Re: Toga II 3.1.2SE

Post by Mike S. »

I guess I need to ignore this version as long as no reasonable, reliable source is provided (I don't mean a source code, but a real source where this version comes from). "Chris Formula" sounds like a joke and I doubt that it is a real life name. Also, the website http://computerchess.org.uk/ seems not to have normal HTML pages even. Any 12 year old can create a site like "Hello, that's me" etc. Why is that missing there?!

All very doubtful and suspicious. I don't like this style. I will not download or use this version as long as I don't see signs of serious origin.
Regards, Mike
User avatar
Graham Banks
Posts: 45587
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Toga II 3.1.2SE

Post by Graham Banks »

Mike S. wrote: Also, the website http://computerchess.org.uk/ seems not to have normal HTML pages even. Any 12 year old can create a site like "Hello, that's me" etc. Why is that missing there?!
Hi Mike,

the CCRL website is http://www.computerchess.org.uk/ccrl/
You left off the important bit on the end.

Regards, Graham.
gbanksnz at gmail.com
User avatar
Eelco de Groot
Posts: 4705
Joined: Sun Mar 12, 2006 2:40 am
Full name:   Eelco de Groot

Re: Toga II 3.1.2SE

Post by Eelco de Groot »

Thanks for the link Graham! Good to see that with this version Chris Formula now seems to fully comply with the GPL license. And interesting that he managed to improve both results for endgames in EET and in Tony Hedlund's positional testsuite. Not bad!

I already made some settings for this new engine, I switched on Toga's extended history pruning as I don't think that is very bad, at least not in previous Toga versions.

Here is the result in r4rk1/p4ppp/qp2p3/b5B1/n1R5/5N2/PP2QPPP/1R4K1 w - - bm Bf6; id "Alekhine-Sterk, Budapest 1921";

The first two Toga 3.2.1 Checkov versions were unsuccessful, but at the third try I got a good result for Alekhine- Sterk after only thirty seconds!

The default version had some problems with this position after finding the solution, going to 18 ply where in the new iteration for Bf6 the engine took more than thirty minutes to calculate a new PV, I did not have enough patience to wait for the next output.

[d]r4rk1/p4ppp/qp2p3/b5B1/n1R5/5N2/PP2QPPP/1R4K1 w - -

Engine: Toga II 3.1.2 Checkov 1b (256 MB)
by Thomas Gaksch and Fabien Letouzey

7/19 0:00 +0.92 1.Nd4 Nc5 2.b4 e5 3.Nf5 f6 4.bxc5 fxg5
5.Qxe5 (46.498)

7/20 0:00 +1.31++ 1.Qf1 f6 2.Be3 Qb5 3.b4 Qe8 4.bxa5 (65.713)

8/20 0:00 +1.32 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.Bxc5 Rxc5 6.Rxc5 bxc5 (82.055)

9/22 0:00 +1.36 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.Ne5 Nb3 6.Rb5 (140.028)

10/26 0:00 +1.32 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.Bd6 Qa3 6.Ne5 h5 (270.237)

11/38 0:01 +1.32 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.h4 f6 6.h5 e5 7.Bxc5 Rxc5
8.Rxc5 bxc5 9.Rb7 (726.091) 660

12/38 0:03 +1.50 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.h3 Qa5 6.Qb1 f6 7.Rh4 f5
8.Qb2 (2.021.665) 666

13/38 0:05 +1.47 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.h3 Qa5 6.Qb1 f6 7.Rh4 f5
8.Qb2 Ne4 (3.699.042) 662

13/40 0:08 +1.70 1.Nd4 Nc5 2.b4 h6 3.Nc6 Qb5 4.Bxh6 Qxc6
5.Qg4 g6 6.Bxf8 Rxf8 7.bxa5 b5 8.Rd4 Rc8 (5.603.855) 663

14/42 0:14 +1.71 1.Nd4 Nc5 2.b4 f6 3.Be3 Bxb4 4.Rbxb4 Rac8
5.Rb1 Rfd8 6.Nb5 Nd7 7.f4 Kh8 8.Rd1 Rxc4
9.Qxc4 (9.960.556) 667

14/44 0:30 +2.10++ 1.Bf6 Rfc8 2.Qe5 Bc3 3.bxc3 Rxc4
4.Qg5 g6 5.Qh6 Rh4 6.Nxh4 Qe2 (20.628.619) 679

15/84 2:24 +3.52++ 1.Bf6 Rfc8 2.Qe5 Rxc4 3.Qg5 Rg4
4.Qxg4 g6 5.Qxa4 Qd3 6.Rc1 Qd6 7.Ne5 Bb4
8.h4 Bd2 9.Qe4 Bxc1 10.Qxa8+ Qf8
11.Qxa7 Bxb2 (99.691.717) 689

16/84 3:19 +3.60 1.Bf6 Rfc8 2.Qe5 Rxc4 3.Qg5 Rg4
4.Qxg4 g6 5.Qxa4 b5 6.Qa3 b4 7.Qa4 Rc8
8.Rd1 Qe2 9.Qd7 Qc2 10.Ne5 (138.425.591) 692

17/84 5:04 +3.66 1.Bf6 Rfc8 2.Qe5 Rxc4 3.Qg5 Rg4
4.Qxg4 g6 5.Qxa4 b5 6.Qa3 b4 7.Qa4 Rc8
8.Rd1 b3 9.Qxb3 Qc4 10.Qxc4 Rxc4
11.Nd4 Rc5 (209.821.293) 689

18/84 7:45 +3.69 1.Bf6 Rfc8 2.Qe5 Rxc4 3.Qg5 Rg4
4.Qxg4 g6 5.Qxa4 b5 6.Qa3 b4 7.Qa4 Qb6
8.Rc1 b3 9.axb3 Bb4 10.Rc6 Qa5
11.Ne5 Rb8 (321.183.406) 689


And for the default settings:


Engine: Toga II 3.1.2SE (256 MB)
by Thomas Gaksch and Fabien Letouzey

7/16 0:00 +0.74 1.Nd4 Nc5 2.b4 h6 3.Be7 Rfe8 4.Bxc5 bxc5
5.Rxc5 Qxe2 6.Nxe2 (34.686)

7/17 0:00 +1.14++ 1.Qf1 f6 2.Be3 Nc3 3.bxc3 Qb7 4.Rg4 (46.479)

8/20 0:00 +1.29 1.Qf1 Nc5 2.b4 Rfc8 3.Be3 Bxb4
4.Rbxb4 Qxa2 5.Bxc5 Rxc5 6.Rxc5 bxc5 (63.218)

9/21 0:00 +1.29 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.Ne5 Nb3 6.Rb5 (129.934)

10/30 0:00 +1.29 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.h4 f6 6.Nd4 e5 7.Bxc5 Rxc5
8.Rxc5 bxc5 (364.950)

11/34 0:01 +1.26 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.h4 f6 6.h5 e5 7.Bxc5 Rxc5
8.Rxc5 bxc5 9.Rb7 (715.808) 620

12/34 0:02 +1.35 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.Kh1 Qa6 6.Ne5 f6 7.Bxc5 fxe5
8.Be3 (1.439.170) 615

13/40 0:07 +1.43 1.Qf1 Nc5 2.b4 Bxb4 3.Rbxb4 Qxa2
4.Be7 Rfc8 5.Bxc5 Rxc5 6.Ra4 Qb2
7.Rcb4 Qc1 8.Rxb6 Rc2 9.Rb7 (4.354.646) 612

14/42 0:19 +1.48 1.Qf1 Nxb2 2.Rxb2 Rac8 3.Rbc2 Rxc4
4.Rxc4 Rc8 5.Rxc8+ Qxc8 6.Qd3 h6
7.Be3 Bc3 8.Qd6 Bf6 9.Bd2 h5 10.h4 b5 (11.752.950) 591

15/48 0:38 +1.56 1.Qf1 Nxb2 2.Rxb2 Rac8 3.Rbc2 Rxc4
4.Qxc4 Qxc4 5.Rxc4 f6 6.Be3 Rd8
7.Kf1 e5 8.Rc7 a6 9.Ra7 e4 10.Nd4 Bc3
11.Rxg7+ Kxg7 12.Ne6+ Kg6 13.Nxd8 (23.896.555) 616

16/48 1:27 +1.60 1.Qf1 Nxb2 2.Rxb2 Rac8 3.Rbc2 Rxc4
4.Qxc4 Qxc4 5.Rxc4 f6 6.Be3 Rd8
7.Kf1 e5 8.Rc7 a6 9.Ke2 g5 10.h3 b5
11.Re7 Bb4 (55.832.828) 640

17/56 2:49 +1.60 1.Qf1 Nxb2 2.Rxb2 Rac8 3.Rbc2 Rxc4
4.Rxc4 Rc8 5.Rd4 Qxf1+ 6.Kxf1 f6
7.Be3 Rc7 8.Rd8+ Kf7 9.Nd4 Bb4
10.Nb5 Rb7 11.Ra8 a5 12.Rd8 e5
13.Ke2 (111.442.427) 656

17/78 6:29 +2.00++ 1.Bf6 Nc5 2.Ne5 gxf6 3.Rg4+ Kh8
4.Qf3 Ne4 5.Rxe4 f5 6.Rh4 f6 7.Nf7+ Rxf7
8.Qxa8+ Kg7 9.Qe8 Qe2 10.Rxh7+ Kxh7
11.Qxf7+ Kh6 (255.832.641) 656

best move: Bg5-f6 time: 34:13.281 min n/s: 672.293 CPU 100.0% n/s(1CPU): 672.293 nodes: 1.380.280.000


An interesting new Toga version!

Eelco
Last edited by Eelco de Groot on Wed Jan 09, 2008 4:19 am, edited 1 time in total.
User avatar
Graham Banks
Posts: 45587
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Toga II 3.1.2SE

Post by Graham Banks »

Mike S. wrote:"Chris Formula" sounds like a joke and I doubt that it is a real life name.
Certainly doesn't sound like a real name, but I'm not 100% certain.
I dislike anonymous authors as much as you do, but at least we have the source code and know what we're dealing with.

Regards, Graham.
gbanksnz at gmail.com
ozziejoe
Posts: 811
Joined: Wed Mar 08, 2006 10:07 pm

Re: Toga II 3.1.2SE

Post by ozziejoe »

great news graham.

when do you reckon the mp version will be available (if ever)?
User avatar
Graham Banks
Posts: 45587
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Toga II 3.1.2SE

Post by Graham Banks »

ozziejoe wrote:great news graham.

when do you reckon the mp version will be available (if ever)?
To be honest I know nothing more than you do. All I did was provide a link to Shaun's post in CCRL public.

Regards, Graham.
gbanksnz at gmail.com
swami
Posts: 6664
Joined: Thu Mar 09, 2006 4:21 am

Re: Toga II 3.1.2SE

Post by swami »

This Toga ahead of Zap Mexico, Naum 2.2, Hiarcs 11.2, Loop 13.6 :shock:
gerold
Posts: 10121
Joined: Thu Mar 09, 2006 12:57 am
Location: van buren,missouri

Re: Toga II 3.1.2SE

Post by gerold »

swami wrote:This Toga ahead of Zap Mexico, Naum 2.2, Hiarcs 11.2, Loop 13.6 :shock:
Not to bad for a clone. :)
User avatar
Graham Banks
Posts: 45587
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Toga II 3.1.2SE

Post by Graham Banks »

Graham Banks wrote:
Mike S. wrote:"Chris Formula" sounds like a joke and I doubt that it is a real life name.
Certainly doesn't sound like a real name, but I'm not 100% certain.
I dislike anonymous authors as much as you do, but at least we have the source code and know what we're dealing with.

Regards, Graham.
I exchanged some polite emails with "Chris Formula". Here are some excerpts:
Chris is my real given name. Formula is my pen name. I'm an avid fan of Toga and Fruit. I live in Manila, Philippines........As far i know, it only requires to release the code to follow the GPL license.

There are so many doubting people in the CCC. As for my case, i just want to help in my own little way to further the rapid development in the computer chess community. The full credits should go to Fabien and Thomas.
I might add that Shaun has been doing some extensive pre-release testing and that this is not another of the recent joke engines. :wink:
gbanksnz at gmail.com