Blitzmonster Toga 3.1.2 SE

[Mike S.]

5m+2s, Athlon@1500(*1)
128 MB hash each
all with Balanced-12.ctg
Single CPU, ponder off
3/4- and R-R-5 piece tbs.
(for some engines)

*1) For this cpu, I used the version for P3 cpus compiled by D.Mendoza which is included in the downloadfile (in the 7-zip archive).

Code: Select all

Toga312-Test 5m+2s, 2008
.
Toga II 3.1.2SE    2915 - Fritz for Fun 4        2740   5.5 - 2.5    +5/-2/=1    68.75%
Toga II 3.1.2SE    2915 - Hiarcs 9               2720   4.0 - 4.0    +4/-4/=0    50.00%
Toga II 3.1.2SE    2915 - Rybka 2.3.2a 32-bit    3000   2.0 - 6.0    +1/-5/=2    25.00%
Toga II 3.1.2SE    2915 - Shredder 9 SE          2790   7.5 - 0.5    +7/-0/=1    93.75%
Toga II 3.1.2SE    2915 - Fruit 51103            2850   6.0 - 2.0    +4/-0/=4    75.00%
Toga II 3.1.2SE    2915 - Bright 0.2c            2750   6.0 - 2.0    +5/-1/=2    75.00%
.
Performance differences from Fritz' table:
.
Toga II 3.1.2SE      2915(*2) -2
Rybka 2.3.2a 32-bit  3000   +105
Hiarcs 9             2720   +195 <--
Fritz for Fun 4      2740    +38
Fruit 51103          2850   -125
Bright 0.2c          2750    -25
Shredder 9 SE        2790   -345 <--
.
*2) The Toga rating means the performance in this test, only. The other ratings are average values between CCRL and CEGT 40/4m.

(Toga without bitbases; Fritz for Fun 4 = 8.1.3.0)

I think this is a great result and a clearly improved Toga again. Thanks to all!

For comparison, see http://tinyurl.com/22rntx (CCRL 40/4m best single versions)

There, currently only Fritz 11, Shredder 11 and Rybka are ranking higher than Toga II 3.1.2 SE. Ranks 4...7 are within 2 Elo points only.

Two game examples with quick wins:

[Event "Toga312-Test 5m+2s"]
[Site "Athlon@1500"]
[Date "2008.01.16"]
[Round "2.5"]
[White "Fruit 51103"]
[Black "Toga II 3.1.2SE"]
[Result "0-1"]
[ECO "A11"]
[WhiteElo "2850"]
[BlackElo "2915"]
[PlyCount "72"]

1. Nf3 d5 2. g3 Bg4 3. Bg2 Nd7 4. c4 e6 5. cxd5 exd5 6. Nc3 c6 7. O-O Ngf6 8.
d3 Bc5 9. h3 Be6 10. Ng5 O-O 11. d4 Bb4 12. Qc2 Re8 13. a3 Bxc3 14. bxc3 h6 15.
Nxe6 Rxe6 16. Rb1 Nb6 17. Bf3 Qd7 18. Qb3 Re7 19. h4 Qh3 20. Qb4 Rae8 21. Qa5
Ne4 22. Bg2 Nc4 23. Qxa7 Qc8 24. Bxe4 b5 25. Qc5 Rxe4 26. Re1 Rxe2 27. Rxe2
Rxe2 28. Bf4 g5 29. hxg5 hxg5 30. Bxg5 Qf5 31. Rf1 Qxg5 32. Qxc6 Nd2 33. Qxb5
Qh5 34. Qb8+ Kg7 35. Qf4 Nxf1 36. g4 Qh3 0-1

[Event "Toga312-Test 5m+2s"]
[Site "Athlon@1500"]
[Date "2008.01.17"]
[Round "8.6"]
[White "Toga II 3.1.2SE"]
[Black "Bright 0.2c"]
[Result "1-0"]
[ECO "D12"]
[WhiteElo "2915"]
[BlackElo "2750"]
[PlyCount "77"]

1. d4 d5 2. c4 c6 3. Nf3 Nf6 4. e3 Bf5 5. Nc3 a6 6. Nh4 Be4 7. f3 Bg6 8. Qb3
Ra7 9. c5 Nbd7 10. Be2 e6 11. g4 Ng8 12. Nxg6 hxg6 13. Bd2 Rh3 14. O-O e5 15.
Rae1 Qg5 16. Rf2 Qd8 17. Bf1 Rh8 18. Rg2 Be7 19. f4 e4 20. Nxe4 dxe4 21. Bc4
Bh4 22. Bxf7+ Kf8 23. Rf1 Ne7 24. Be6 Qe8 25. Ba5 Nf6 26. Bc7 a5 27. Bd6 Qd8
28. Qc2 a4 29. Be5 a3 30. g5 Nfd5 31. Qxe4 Nc7 32. f5 Nxe6 33. fxe6+ Kg8 34.
Rf7 Rh7 35. Bd6 b5 36. cxb6 Rb7 37. Bxe7 Rxe7 38. b7 Rxf7 39. exf7+ 1-0

[Jouni]

Yes new Toga is surprisingly good blitzer. Here's match against Fritz (10 moves book, Pentium 2400):

Code: Select all

                     
1   Toga II 3.1.2SE  ½½½01111½½10½1½½½00010111½½½11½000½0½½½½1½1½011½½½  27.5/50
2   Fritz 10         ½½½10000½½01½0½½½11101000½½½00½111½1½½½½0½0½100½½½  22.5/50

But what's Toga's secret? In all test suites Fritz is clearly better (specially in tactics). Is it "positional" play?

Jouni

[perejaslav]

Toga Checkov Beta (3.2.1 SE) based on Fruit 2.1 testing on Q6600@3500

Code: Select all

Toga II 3.2.1SE - HIARCS 11.1 UCI            50.0    -    50.0        50.00%        
Toga II 3.2.1SE - Loop 13.6        44.0    -    56.0        44.00%                        
Toga II 3.2.1SE - Zappa Mexico        55.5    -    44.5        55.50%
Toga II 3.2.1SE - Spike 1.2 Turin        65.5    -    34.5        65.50%
Toga II 3.2.1SE - Fruit 2.3                        47.5    -    52.5        47.50%        
Toga II 3.2.1SE - Toga II 3.1.2SE        39.5    -    60.5        39.50%    
Toga II 3.2.1SE - Rybka 2.3.2a 32-bit        25.0    -    74.0        25.25%        
Toga II 3.2.1SE - Shredder 11 UCI        37.5    -    61.5        37.88%
Toga II 3.2.1SE - Naum 2.2        51.5    -    48.5        51.50%
Toga II 3.2.1SE - Ktulu 8        59.0    -    41.0        59.00%

Code: Select all

Toga II 3.2.1SE - Belka 2.0.0         44.0    -    56.0        44.00%
Toga II 3.2.1SE - Strelka 2.0 B       44.5    -    55.5        44.50%
Toga II 3.2.1SE - Belka 1.8.18        46.0    -    54.0        46.00%
Toga II 3.2.1SE - Belka 1.8.16        40.0    -    60.0        40.00%
Toga II 3.2.1SE - Belka 1.8.18-F5     42.5    -    57.5        42.50%

Code: Select all

1   Fritz 11         +41/-23/=36 59.00%   59.0/100
2   Toga II 3.2.1SE  +23/-41/=36 41.00%   41.0/100

[Jouni]

BTW why this "3.1.2" name -isn't "1.3.2" more logical?

Jouni

[Eelco de Groot]

Toga Checkov Beta (3.2.1 SE) based on Fruit 2.1 testing on Q6600@3500

Code: Select all

Toga II 3.2.1SE - HIARCS 11.1 UCI            50.0    -    50.0        50.00%        
Toga II 3.2.1SE - Loop 13.6        44.0    -    56.0        44.00%                        
Toga II 3.2.1SE - Zappa Mexico        55.5    -    44.5        55.50%
Toga II 3.2.1SE - Spike 1.2 Turin        65.5    -    34.5        65.50%
Toga II 3.2.1SE - Fruit 2.3                        47.5    -    52.5        47.50%        
Toga II 3.2.1SE - Toga II 3.1.2SE        39.5    -    60.5        39.50%    
Toga II 3.2.1SE - Rybka 2.3.2a 32-bit        25.0    -    74.0        25.25%        
Toga II 3.2.1SE - Shredder 11 UCI        37.5    -    61.5        37.88%
Toga II 3.2.1SE - Naum 2.2        51.5    -    48.5        51.50%
Toga II 3.2.1SE - Ktulu 8        59.0    -    41.0        59.00%

Code: Select all

Toga II 3.2.1SE - Belka 2.0.0         44.0    -    56.0        44.00%
Toga II 3.2.1SE - Strelka 2.0 B       44.5    -    55.5        44.50%
Toga II 3.2.1SE - Belka 1.8.18        46.0    -    54.0        46.00%
Toga II 3.2.1SE - Belka 1.8.16        40.0    -    60.0        40.00%
Toga II 3.2.1SE - Belka 1.8.18-F5     42.5    -    57.5        42.50%

Code: Select all

1   Fritz 11         +41/-23/=36 59.00%   59.0/100
2   Toga II 3.2.1SE  +23/-41/=36 41.00%   41.0/100

Thanks a lot for doing this test Slavik! I finally can get an idea now how good Toga Checkov plays against other programs. Apart from Spike, Hiarcs 11.1, Toga of course and Fruit 2.3 I have none of them myself so it is very useful.

Strelka and Belka did very well! The space dogs and squirrels are good opponents, tough for Toga!

If I try to calculate some TPR, from the first ten matches I get something like this using the ratings of the CCRL 40/4 list:

Toga Checkov Beta 3 scores 475 out of 1000 games.

Ratings of the opponents:

Rybka 2.3.2a 32-bit 3025
Hiarcs 11.1 2908
Loop 13.6 32-bit 2874
Zappa Mexico 64-bit 2907
Spike 1.2 Turin 2826
Fruit 2.3 2876
Toga II 3.1.2SE ca. 2915?
Shredder 11 2972
Naum 2.2 32-bit 2881
Ktulu 8.0 2827

Average rating of the opponents = 2901

Toga Checkov scores 47.5 % that gives a TPR of 2901 - (2.5 * 7) ≈ 2884

I am happy Beating Zappa Mexico in your test especially seems quite a feat, and a draw against Hiarcs 11.1 too, I have not yet taken the result against Fritz 11 into account for the elo calculation, I am very happy with those results too, even against Shredder 11 not so bad, I would like to own Shredder 11, seems a good program to test against. The result against Loop could clearly be better, and overall there are quite a few surprising results, not uncommon I suppose but it clearly shows every engine has its own weaknesses. Rybka 2.3.2a still reigns supreme! The Russian programs are difficult to beat!

My thanks!
Eelco

P.S. The 3.2.1 SE designation was a mistake in the read.me, that should have been 3.1.2SE really because the code from Charles' version was the basis for Toga Checkov Beta 3. I got confused with all those numbers!

[Philippe]

There are currently so many versions and compiles of Toga 3.1.2 SE versions, can you precise which one you tested ? Giving the size of the exe, its date and where you downloaded it from. Thanks.

[Mike S.]

I used the version for P3 cpus compiled by D.Mendoza which is included in the downloadfile from CCRL, in the 7-zip(!)* archive included in the zip file. I assume only people using an old cpu will be interested in that version, but the P4 versions didn't run on that computer with AMD AthlonXP cpu.

Code: Select all

2007-12-19  19:43           286.720 Toga II 3.1.2SEp3.exe
               1 Datei(en),        286.720 Bytes

http://computerchess.org.uk/Toga/TogaIIv3.1.2SE.zip

*) 7-zip is another file compression format. You need an unpacker which supports it. See http://www.7-zip.org/

[Denis P. Mendoza]

I've added some compiles for much older pc (as per request by some collegues) for Pentium 1 and Pentium 2.

http://kirr.homeunix.org/chess/discussi ... hp?id=8971

The compiles were officially released on this thread at CCRL forum:

http://kirr.homeunix.org/chess/discussi ... 085acbfb0f

FYI.

[gerold]

There are currently so many versions and compiles of Toga 3.1.2 SE versions, can you precise which one you tested ? Giving the size of the exe, its date and where you downloaded it from. Thanks.

2/12 blitz hp dual core.

Toga 3.1.2-Fruit 2.3.1 26.5/53 - 26.5/53.

Rybka 1.0 -Toga 3.21 12/23 - 11/23.

Fruit 2.3.1- Rybka 1.0 90/157 - 67/157

Toga is very strong now as for as the free engines.

In my test Fruit 2.3.1 is the strongest free engine.

Best to you,

Gerold.