How strong are TSCP and Faile?

[Tord Romstad]

I am currently working on implementing the UCI_Elo UCI option in Glaurung. In order to calibrate the ratings, I want to run some matches against weak engines. Does anybody know the approximate ratings for TSCP and Faile on a 2 GHz computer? Are there any open source programs even weaker than TSCP?

Tord

[Peter Eizenhammer]

Hi Tord,

for Faile I have some data (dont stop reading before end of table .)
tscp might be (even) some weaker, but i dont know yet.

Code: Select all

01: Fruit20              102,0/121
02: Glaurung021          100,0/121
03: craftyClassic2004uci 94,0/121 
04: Pepito159            92,0/121 
05: Slow296              90,5/121 
06: Amy-net-087          79,0/121 
07: phalanxXXII          77,5/121 
08: Arasan83             66,5/121 
09: GnuChess-507-Net     61,5/121 
10: GES_134              60,0/121 
11: Gnuchess4TM          56,5/121 
12: ThorsHammer225       53,0/121 
13: Natwarlal009         52,5/121 
14: Beowulf 2.3          48,5/121 
15: GreKo30              43,0/121 
16: Olithink413          42,5/121 
17: Sjeng112             33,0/121 
18: Parrot050205         33,0/121 
19: SunsetterC7          33,0/121 
20: GERBIL02             29,5/121 
21: faile14              21,0/121 


Peter

[Dan Honeycutt]

Are there any open source programs even weaker than TSCP?

Simon. See, ie:

http://volker-pittlik.name/wbforum/viewtopic.php?t=1735

Dan H.

[Michael Yee]

Hi Tord,

Here are a few that are weaker or about the same level as TSCP:

- simon
- xadreco
- nero (pascal)
- pyotr (java w/o source)
- jester
- mscp?
- witz

There are some other java engines, but I don't know how stable they are (eden, tiffany, etc.).

Some other ideas for testing against weaker opponents:

- beowulf has a strength parameter
- phalanx has option : "-e <easy level 0...100> default: 0 (best play)"
- craftyclassic has intensity and other "special edition" ways to limit strength

I'm glad to hear that you're thinking about limiting strength. I'm also interesting in simulating human-like play (esp. mistakes)... Maybe a discussion could be started in a new thread. Also, I recall that some people in the CCC forum were very interested in this, too, for youth chess instruction (Odd Gunner Malin? Guenther Simon? or somebody else--I don't remember exactly).

Michael

[Uri Blass]

Tord,Nero is open source program that is clearly weaker than tscp
It is not written in C but you can download source at

http://www.mit.jyu.fi/~huikari/download.html

I think that there are other open source program that are weaker than tscp.

Uri

[Pallav Nawani]

Hi,

These are approximate numbers, except for Faile, which is quite accurate. All open source engines.

TSCP: ~1800
Faile: ~2100
Gnuchess5 ~2200
Sjeng 11.2 ~2300
Beowulf ~2350

Pallav

[Dann Corbit]

I am currently working on implementing the UCI_Elo UCI option in Glaurung. In order to calibrate the ratings, I want to run some matches against weak engines. Does anybody know the approximate ratings for TSCP and Faile on a 2 GHz computer? Are there any open source programs even weaker than TSCP?

Tord

Here is a list at slow time control of older versions of engines.
It is calibrated at G/90 or slower on Pentium II 300 MHz of time.

Code: Select all

   Program            Elo    +   -   Games   Score   Av.Op.  Draws
 1 Crafty-18.10     : 2646   27 106   247    88.5 %   2292   10.9 %
 2 Yace 0.99.50     : 2597   29  91   228    85.1 %   2294   13.2 %
 3 LG2000V3         : 2589   58 148    70    80.7 %   2341    7.1 %
 4 Comet-B37        : 2585   28  85   237    84.0 %   2298   13.5 %
 5 ZChess-222       : 2576   31  94   197    84.3 %   2285   13.2 %
 6 TCBishop-4601    : 2566   57 135    71    80.3 %   2322   11.3 %
 7 MAD-005          : 2564   29  85   242    82.2 %   2298   10.7 %
 8 Amy-07           : 2554   32  99   196    83.7 %   2270   10.2 %
 9 Phalanx-xxii     : 2542   59 124    74    75.7 %   2344    8.1 %
10 AnMon-510        : 2540   29  81   239    81.0 %   2289   11.3 %
11 Gromit3          : 2536   59  91    74    75.0 %   2346   25.7 %
12 AnMon-509        : 2535   30  74   234    80.3 %   2290   15.4 %
13 Nejmet-260       : 2532   61 117    71    75.4 %   2338   12.7 %
14 Gromit2          : 2524   63  91    68    73.5 %   2347   26.5 %
15 Quark-146a       : 2511   61 109    72    73.6 %   2332   13.9 %
16 TCBishop-0045    : 2510   62 103    70    73.6 %   2332   18.6 %
17 VCTAO44          : 2483   65 105    69    70.3 %   2333   13.0 %
18 GLC-213          : 2478   31  65   240    74.8 %   2289   12.9 %
19 ZChess-120       : 2475   32  63   231    74.2 %   2291   15.2 %
20 FranWB-090       : 2466   32  61   242    72.9 %   2294   14.5 %
21 Yace-020         : 2446   34  59   229    70.5 %   2294   14.4 %
22 WildCat-257      : 2444   33  55   237    71.1 %   2287   18.1 %
23 ExChess-402      : 2438   69  83    74    62.8 %   2347   17.6 %
24 Arasanx-54       : 2435   34  57   242    69.0 %   2296   12.4 %
25 Pepito-121       : 2425   71  85    71    62.0 %   2340   16.9 %
26 FRANWB-081       : 2422   34  55   247    67.8 %   2292   11.7 %
27 Ant-606          : 2417   35  52   241    66.4 %   2299   14.9 %
28 Ant-416          : 2384   37  49   239    62.6 %   2295   13.8 %
29 Bringer-18       : 2375   40  49   218    60.3 %   2303   14.2 %
30 Amyan-1423       : 2370   37  47   252    61.3 %   2290   13.1 %
31 Knightx-154      : 2369   79  74    74    53.4 %   2346   14.9 %
32 ExChess-311      : 2367   80  80    71    50.0 %   2367    9.9 %
33 GNUChess-414     : 2360   82  79    70    52.9 %   2340   11.4 %
34 InmiChess3       : 2359  105 122    35    48.6 %   2368   17.1 %
35 kke-196          : 2348   87  83    62    53.2 %   2325   12.9 %
36 chezzz-088       : 2345   40  46   243    56.8 %   2298    9.9 %
37 Sjeng10cct       : 2344   81  74    72    52.1 %   2329   15.3 %
38 Dragon-311       : 2331   77  80    69    45.7 %   2361   15.9 %
39 LDBlanche-20c    : 2321   73  78    75    46.7 %   2345   16.0 %
40 Betsy-526        : 2312   43  42   243    51.9 %   2299   10.7 %
41 Ghost-13         : 2308   43  40   242    52.3 %   2292   14.5 %
42 OliThink-305c    : 2305   86  78    74    45.9 %   2333    2.7 %
43 Fortress-162     : 2300   45  37   241    50.4 %   2297   20.3 %
44 GNU Chess v5.02  : 2280   84  76    68    41.2 %   2342   14.7 %
45 WBQueen-214      : 2270   82  72    73    40.4 %   2338   15.1 %
46 Bestia-07        : 2269   43  42   244    46.3 %   2295   10.7 %
47 Sjeng73          : 2262   80  72    74    40.5 %   2329   16.2 %
48 Cilian-413       : 2256   37  41   238    44.3 %   2296   27.3 %
49 Amyan-111        : 2254   44  40   244    43.2 %   2301   13.5 %
50 Ufim143          : 2243   71  58   102    36.8 %   2337   18.6 %
51 Gully213p16      : 2239   97  68    73    35.6 %   2342    8.2 %
52 Gaviota-022      : 2238   46  40   235    41.5 %   2297   13.2 %
53 Faile-14         : 2237   44  40   238    41.8 %   2295   15.5 %
54 Averno-32        : 2236   50  42   221    42.5 %   2288    6.3 %
55 SSEchess-202     : 2227  108  69    68    33.8 %   2343    5.9 %
56 SSEchess-2045    : 2215  103  67    72    34.0 %   2330    6.9 %
57 DrChess-20       : 2208  101  64    70    29.3 %   2361   15.7 %
58 Noonian-30       : 2196   98  64    74    31.1 %   2334   13.5 %
59 Morphy-322       : 2191  109  62    74    29.1 %   2346    9.5 %
60 Chessterfield-15a: 2185   53  35   243    33.5 %   2304   12.8 %
61 Esc-028          : 2161  140  63    65    23.1 %   2370    6.2 %
62 Freyr-1067       : 2143   67  33   242    28.3 %   2304    5.4 %
63 Grizzly-125      : 2142  134  59    73    23.3 %   2349    5.5 %
64 LarsenVB-407     : 2120  142  60    69    21.7 %   2343    5.8 %
65 Aldebaran-070    : 2117   67  32   231    25.3 %   2305   12.6 %
66 TSCP-152         : 2116  117  59    70    21.4 %   2342   17.1 %
67 Beowulf-16b      : 2114  143  70    50    20.0 %   2355   16.0 %
68 Gerbil-01        : 2087   75  30   247    22.7 %   2300    8.1 %
69 Storm-006        : 2076  162  59    67    18.7 %   2331    4.5 %
70 TSCP-172         : 2069  160  59    66    18.2 %   2331    6.1 %
71 DChess-102       : 2061   88  29   247    20.0 %   2301    4.5 %
72 31337-30         : 2057   88  29   240    19.0 %   2310    7.1 %
73 EnginMax-260     : 2049  154  57    66    14.4 %   2358   13.6 %
74 LordKing-29D     : 2038  170  54    72    13.9 %   2355    8.3 %
75 Mint v2.3        : 2035  169  56    69    15.2 %   2333    7.2 %
76 LordKing-29      : 2024  158  51    79    14.6 %   2332    8.9 %
77 Delfi 1.02       : 2021  186  49    81    12.3 %   2362    4.9 %
78 Blikskottel-05   : 2013  108  28   236    15.7 %   2305    3.4 %
79 Noonian-29       : 1970  199  52    71    10.6 %   2341    7.0 %
80 LarsenVB-313     : 1954  199  51    74     9.5 %   2346    8.1 %
81 Mustang-172      : 1954  247  53    69     9.4 %   2347    1.4 %
82 Ozwald-043       : 1945  231  51    72     9.7 %   2332    2.8 %
83 Golem01          : 1919  244  52    68     8.1 %   2341    4.4 %
84 MSCP-011         : 1874  289  52    66     6.1 %   2350    3.0 %
85 NeroWB-5         : 1845  295  49    74     5.4 %   2343    2.7 %
86 Raffaela-005     : 1765  367  33    68     3.7 %   2332    1.5 %
87 Tamerlane-01     : 1731  268   0    67     0.7 %   2331    1.5 %
88 ChadChess-015    : 1710  271   0   220     2.7 %   2310    1.8 %


[Uri Blass]

I see

66 TSCP-152 : 2116 117 59 70 21.4 % 2342 17.1 %
67 Beowulf-16b : 2114 143 70 50 20.0 % 2355 16.0 %
68 Gerbil-01 : 2087 75 30 247 22.7 % 2300 8.1 %

I do not believe that a single version of gerbil was weaker than tscp.

I believe that if engines do not score between 40-60% the rating can be misleading and the problem is not only statistical error.

When you do not care to give every player except the weaker players and the strongest players opponents that they score near than 50%your rating is simply misleading.

on the other hand if you do like Igor does in the infinite loop and give players to play almost only against players with similiar strength your rating is also misleading.

Your rating have difference that is too small when Igor's rating in the infinite loop has difference that is too high.

Uri

[Ciro Vignotto]

Does anybody know the approximate ratings for TSCP and Faile on a 2 GHz computer? Are there any open source programs even weaker than TSCP?

Tord

Code: Select all

Faile 1.4.4                 2153
TSCP 1.81                   2033

Elo rating from my DB (over 180.000 games).
http://vigo.altervista.org/ -> rating list
HTH, Ciro

[Michael Yee]

Another thought... You can try to estimate the human-calibrated rating of TSCP (and others) in an indirect way by having someone with a windows machine and chessmaster run a couple tournaments against some personalities. (You could also throw in delfi's weakened players too)

What I envision is a tournament with:

cm 1200
cm 1300
cm 1400
cm 1500
cm 1600
cm 1700
cm 1800
cm 1900
cm 2000
delfi 1250
delfi 1300
delfi 1400
delfi 1500
delfi 1600
delfi 1700
tscp
faile
mscp
etc

Michael

[Pedro Castro]

Nobody has commented the ELO that it indicates to that time control corresponds, I believe that it is important.

On the other hand to comment my experience, in the list appears my program Danasah 1.1 ELO 2006, I believe that it is a ELO guessed right in games of 5 minutes against other programs.

But when he plays with a human (2150), the human does not have opportunity to even gain in games 5 minutes and in games of 25 games also (plus time not proven).

A Grand Master who played with Danasah (Mario Gomez 2470) indicated that its level of game was superior to 2200 points, he played 3 games and he lost one (GM ~ 10 minutes per time operator - DanaSah 5 minutes).

My rating ELO is 1830 and I cannot win I believe TSCP, Danasah 1.1 with book I believe that he is similar to TSCP 1.81 in games 5 minutes, so I share the ELO given by Vito.

[Tord Romstad]

Hi all,

Thanks for the replies! I am playing a bullet match with Glaurung set at 1800 Elo (slowdown by a factor of approximately 50 compared to Glaurung at full strength) against TSCP right now. After 73 games, Glaurung is leading by 48.5-24.5, which puts it about 100 Elo points ahead of TSCP. Seems like I am not too far off the mark, but that I should perhaps tune my ratings up a tiny bit.

I think I'll try Simon next (if I am able to compile it in Mac OS X). Dan, do you have any estimate of how much weaker it is compared to TSCP?

If somebody would like to test the new Glaurung development version with experimental UCI_LimitStrength, feel free to drop me an e-mail or a personal message at this board, and I will send you the program.

Tord

[Dann Corbit]

I see

66 TSCP-152 : 2116 117 59 70 21.4 % 2342 17.1 %
67 Beowulf-16b : 2114 143 70 50 20.0 % 2355 16.0 %
68 Gerbil-01 : 2087 75 30 247 22.7 % 2300 8.1 %

I do not believe that a single version of gerbil was weaker than tscp.

I believe that if engines do not score between 40-60% the rating can be misleading and the problem is not only statistical error.

When you do not care to give every player except the weaker players and the strongest players opponents that they score near than 50%your rating is simply misleading.

on the other hand if you do like Igor does in the infinite loop and give players to play almost only against players with similiar strength your rating is also misleading.

Your rating have difference that is too small when Igor's rating in the infinite loop has difference that is too high.

Uri

The raw data does not show that TSCP is stronger. There is considerable overlap in the error bars. Also, Beowulf has lots of failures because of running on machines that did not support some particular inline assembly (which will show up in the PGN files as 'lost on time') and so the Beowulf rating is probably particularly onerous.

The rating above is just a dump of the raw data on data that is still generating in a long running contest. But there are over 6000 games and so for the most part, I think the answers will be as good or better than most other contests.

Keep in mind that the answer is only for games at this time control.

TSCP did well in my previous contest "Battle of the Crowns" also:

Code: Select all

    Program          Score     %    Av.Op.  Elo    +   -    Draws

  1 TSCP         :  30.0/ 36  83.3   2275   2555   83 155   22.2 %
  2 Monik        :  25.0/ 36  69.4   2289   2432   96 146   11.1 %
  3 Storm        :  21.5/ 36  59.7   2297   2365  107 112   19.4 %
  4 Noonian      :  21.0/ 36  58.3   2298   2356  109  78   44.4 %
  5 Ozwald       :  19.5/ 36  54.2   2300   2330  115 104   19.4 %
  6 Zephyr       :  19.5/ 36  54.2   2300   2330  115  97   25.0 %
  7 LarsenVB     :  19.5/ 36  54.2   2300   2330  115  97   25.0 %
  8 SnailSCP     :  10.0/ 36  27.8   2320   2154  129  93   22.2 %
  9 Raffaela     :   9.0/ 36  25.0   2322   2132  135  90   22.2 %
 10 Golem01      :   5.0/ 36  13.9   2335   2018  237  81    5.6 %


(the Elo rankings are only relatvie, of course)

I think that at long time control, TSCP is stronger than you think.

[Dann Corbit]

Code: Select all

[Event "Computer chess game"]
[Site "LAZARUS"]
[Date "2002.02.09"]
[Round "1"]
[White "Amyan-1423"]
[Black "TSCP-152"]
[Result "1/2-1/2"]
[ECO "C55a"]
[Variation "Two Knights Defence"]
[TimeControl "21600"]

1.e4 Nc6 2.Nf3 e5 3.Bc4 Nf6 4.O-O Nxe4 5.d3 Nf6 6.Nc3 d6 7.Ng5 d5 8.Nxd5
Nxd5 9.Nxf7 Kxf7 10.Qf3+ Ke6 11.Be3 Be7 12.Rfe1 a6 13.Bd2 Bc5 14.Bf4 Rf8
15.d4 Bxd4 16.Rad1 Rf5 17.Qe4 Kf6 18.Bg3 Nb6 19.Bb3 Bd7 20.c3 Bc5 21.Bc2
g6 22.b4 Be7 23.Qe3 Rh5 24.Be4 Qc8 25.Bf3 Bg4 26.Bxc6 Bxd1 27.Be4 Bg4 28.
h3 Nc4 29.Qd3 Nb2 30.Qe3 Nc4 31.Qd3 Nb2 32.Qe3 Nc4 1/2-1/2

[Event "Computer chess game"]
[Site "LAZARUS"]
[Date "2002.02.10"]
[Round "2"]
[White "TSCP-172"]
[Black "Amyan-1423"]
[Result "1/2-1/2"]
[ECO "C50j"]
[Variation "Giuoco Piano: Four Knights Variation"]
[TimeControl "21600"]

1.e4 e5 2.Nf3 Nf6 3.Nc3 Nc6 4.Bc4 Bc5 5.Ng5 O-O 6.d3 d6 7.Nxf7 Rxf7 8.
Bxf7+ Kxf7 9.Bg5 h6 10.Bxf6 Qxf6 11.O-O Be6 12.Na4 Bb6 13.c3 Kg8 14.b4 Qg5
15.Nxb6 axb6 16.a4 Rf8 17.Qc2 Bh3 18.f3 Bxg2 19.Qxg2 Qe3+ 20.Qf2 Qxd3 21.
Rac1 Kh8 22.Qg3 b5 23.axb5 Qxb5 24.f4 exf4 25.Rxf4 Rxf4 26.Qxf4 Ne5 27.
Qf8+ Kh7 28.Qf5+ g6 29.Qf6 Qb6+ 30.Kg2 Qc6 31.Qe7+ Kg8 32.Re1 Qc4 33.Re3
Qa2+ 34.Kg3 Qc4 35.Qh4 Kg7 36.Qd8 Qf7 37.h3 h5 38.h4 Qf1 39.Qe7+ Nf7 40.
Qd7 Qg1+ 41.Kf3 c6 42.Qh3 g5 43.Ke2 g4 44.Qg3 Qh1 45.Kd2 Ne5 46.Kc2 Qf1
47.Re1 Qc4 48.Qe3 Qa2+ 49.Kc1 Kh7 50.Re2 Qa1+ 51.Kc2 Qa4+ 52.Kc1 Qa3+ 53.
Kb1 Qb3+ 54.Kc1 Qa4 55.Rf2 Nc4 56.Qe2 Qa1+ 57.Kc2 Qb2+ 58.Kd3 Ne5+ 59.Kd4
Qa1 60.Qa2 Qd1+ 61.Ke3 Qd3+ 62.Kf4 g3 63.Kg5 gxf2 64.Qxf2 Nf3+ 65.Kf4 Nd2
66.Qe3 Qc2 67.Qe2 Qxc3 68.Qxh5+ Kg7 69.Qg4+ Kf6 70.Qg5+ Kf7 71.Qf5+ Ke7
72.Qg5+ Kd7 73.Qf5+ Kc7 74.Qa5+ b6 75.Qa7+ Kc8 76.Qa8+ Kd7 77.Qb7+ Ke6 78.
Qc8+ Ke7 79.Qc7+ Ke8 80.Qc8+ Kf7 81.Qd7+ Kg6 82.Qf5+ Kh6 83.Qg5+ Kh7 84.
Qf5+ Kg8 85.Qc8+ Kf7 86.Qd7+ Kg6 87.Qf5+ Kh6 88.Qg5+ Kh7 89.Qf5+ Kg8 90.
Qc8+ Kf7 1/2-1/2

[Event "Computer chess game"]
[Site "LAZARUS"]
[Date "2002.10.26"]
[Round "2"]
[White "TSCP-152"]
[Black "Ant-606"]
[Result "1/2-1/2"]
[ECO "B30p"]
[Variation "Sicilian: Rossolimo, 3...e6 4.Nc3"]
[TimeControl "21600"]

1.e4 c5 2.Nf3 e6 3.Nc3 Nc6 4.Bb5 Nd4 5.O-O Nxb5 6.Nxb5 a6 7.Nc3 b5 8.d4
cxd4 9.Qxd4 Bb7 10.Be3 Qc7 11.Rfd1 Nf6 12.e5 Nd5 13.Nxd5 Bxd5 14.Ne1 Be7
15.a4 O-O 16.axb5 axb5 17.Ra7 Qxa7 18.Qxa7 Rxa7 19.Bxa7 d6 20.exd6 Bxd6
21.Bd4 Rc8 22.g3 f5 23.f4 h6 24.Kf2 g5 25.Ke3 Kf7 26.Ra1 gxf4+ 27.gxf4 Rg8
28.Ra7+ Ke8 29.Rg7 Rxg7 30.Bxg7 h5 31.c3 Be4 32.Bd4 Kd7 33.b3 Bd5 34.b4
Be4 35.Be5 Be7 36.Kd4 Kc6 37.Nd3 Bxd3 38.Kxd3 Kd5 39.Ke3 Bh4 40.Kf3 Bd8
41.Ke3 Be7 42.Kd3 Bf8 43.Ke3 Kc4 44.Kd2 Be7 45.Ke3 Bd8 46.Kd2 Bb6 47.Ke2
Bg1 48.h3 Ba7 49.Kd2 Bb6 50.Bb8 Bd8 51.Be5 Bh4 52.Bd4 Be7 53.Be5 Bd8 54.
Ke2 Bb6 55.Kd2 Kd5 56.Kd3 Bd8 57.Ke3 Be7 58.Kd3 Bh4 59.Bd4 Bd8 60.Be5 Be7
61.Ke3 Bh4 62.Bd4 Bg3 63.Be5 Bh2 64.Kd3 Bg1 65.Bc7 Ba7 66.Be5 Bb6 67.Bb8
Bf2 68.Be5 Ba7 69.Bc7 Bg1 70.Be5 Bh2 71.Bc7 Bg3 72.Be5 Bf2 73.Bc7 Be1 74.
Be5 h4 75.Ke2 Bg3 76.Kd3 Bf2 77.Bf6 Be1 78.Bd4 Kd6 79.c4 Bxb4 80.cxb5 Kd5
81.b6 Kc6 82.Bf2 Bd6 83.Be3 Be7 84.Kc4 Ba3 85.Kb3 Bf8 86.Kc4 Be7 87.Bd4
Ba3 88.Be3 Bf8 89.Kd3 Bh6 90.Kd2 Bg7 91.Bf2 Bf6 92.Ke2 Be7 93.Kd3 Bd8 94.
Kc4 Bxb6 95.Bxh4 Kd6 96.Bf6 Bf2 97.h4 Bxh4 98.Bxh4 e5 99.fxe5+ Kxe5 100.
Kd3 Kf4 101.Kd4 Kg4 102.Be1 f4 103.Ke4 Kg5 104.Bd2 Kh6 105.Kxf4 Kh5 106.
Kf5 Kh4 107.Kf4 Kh3 108.Bc3 Kh2 109.Kf3 Kh1 110.Bg7 Kg1 111.Be5 Kh1 112.
Bf6 Kg1 113.Be5 Kh1 114.Bf6 Kg1 115.Be5 1/2-1/2

[Event "Computer chess game"]
[Site "LAZARUS"]
[Date "2003.12.06"]
[Round "1"]
[White "Bringer-18"]
[Black "TSCP-152"]
[Result "0-1"]
[ECO "C65k"]
[Variation "Spanish: Berlin, 4.O-O"]
[TimeControl "21600"]

1.e4 Nc6 2.Nf3 e5 3.Bb5 Nf6 4.O-O a6 5.Bxc6 dxc6 6.Nxe5 Bd6 7.Nf3 Bg4 8.e5
Bxf3 9.Qxf3 Bxe5 10.Re1 Qd6 11.Qf4 Nd7 12.d4 O-O 0-1

[Event "Computer chess game"]
[Site "LAZARUS"]
[Date "2004.07.18"]
[Round "1"]
[White "Cilian-413"]
[Black "TSCP-152"]
[Result "1/2-1/2"]
[ECO "D15t"]
[Variation "Slav: Geller Gambit"]
[TimeControl "21600"]

1.d4 d5 2.c4 dxc4 3.Nf3 Nf6 4.Nc3 c6 5.e4 b5 6.a3 e6 7.Be2 Be7 8.O-O O-O
9.Bg5 Nbd7 10.Bf4 a5 11.Ng5 Bb7 12.Qc2 Nh5 13.Nxh7 Nxf4 14.Nxf8 Qxf8 15.
Bg4 b4 16.axb4 axb4 17.Ne2 Rxa1 18.Rxa1 Nd3 19.Bf3 e5 20.Qxc4 Nxb2 21.Qb3
exd4 22.Qxb2 c5 23.Ng3 d3 24.Ra5 Bd6 25.Qa1 Ne5 26.Bd1 Qb8 27.Nf5 Bxe4 28.
f3 Bc7 29.fxe4 Bxa5 30.Qxa5 b3 31.Ne7+ Kh7 32.Qe1 g6 33.Bxb3 Qxb3 34.h3
Qb2 35.Nd5 d2 36.Qh4+ Kg7 37.Qf6+ Kh6 38.Qh8+ Kg5 39.Qd8+ Kh6 40.Qh8+ Kg5
41.Qf6+ Kh6 1/2-1/2

[Event "Computer chess game"]
[Site "LAZARUS"]
[Date "2004.07.19"]
[Round "2"]
[White "TSCP-152"]
[Black "Cilian-413"]
[Result "1/2-1/2"]
[ECO "C46"]
[Variation "Three Knights: 3...Bc5 4.Nxe5"]
[TimeControl "21600"]

1.e4 e5 2.Nc3 Nc6 3.Nf3 Bc5 4.Nxe5 Bxf2+ 5.Kxf2 Nxe5 6.d4 Ng6 7.Bb5 c6 8.
Be2 Qh4+ 9.Kg1 d6 10.d5 Bd7 11.Be3 Qe7 12.Qf1 Qe5 13.dxc6 Bxc6 14.Qf2 Qa5
15.Bd4 Nf6 16.Bxf6 gxf6 17.Bc4 Ne5 18.Bd5 O-O-O 19.Rf1 Bxd5 20.exd5 Rhg8
21.a4 Ng4 22.Qd4 Qc5 23.Qxc5+ dxc5 24.Re1 Rge8 25.Rf1 Rd6 26.Rf3 Re5 27.h3
Re1+ 28.Rf1 Rxf1+ 29.Kxf1 Ne3+ 30.Kf2 Nxc2 31.g3 b6 32.Kf3 f5 33.Kf4 Nd4
34.Re1 Kd7 35.Ra1 Rh6 36.Rh1 a6 37.Kg5 Rg6+ 38.Kf4 b5 39.Re1 Rh6 40.axb5
axb5 41.Rh1 Kd6 42.Ra1 Rxh3 43.Ra6+ Kd7 44.Ra7+ Kd6 45.Ra6+ Kd7 46.Ra7+
Kd6 47.Ra6+ 1/2-1/2

[Event "Computer chess game"]
[Site "LAZARUS"]
[Date "2004.07.19"]
[Round "1"]
[White "Cilian-413"]
[Black "TSCP-172"]
[Result "0-1"]
[ECO "D15t"]
[Variation "Slav: Geller Gambit"]
[TimeControl "21600"]

1.d4 d5 2.c4 dxc4 3.Nf3 Nf6 4.Nc3 c6 5.e4 b5 6.a3 e6 7.Be2 Be7 8.O-O O-O
9.Bg5 Nbd7 10.Bf4 a5 11.Ng5 Bb7 12.Qc2 h6 13.Nf3 b4 14.Na4 bxa3 15.bxa3 c5
16.e5 Nd5 17.Bg3 cxd4 18.Nxd4 N7b6 19.Nxb6 Nxb6 20.Nf3 Rc8 21.Rfd1 Qc7 22.
Nd4 Rfd8 23.Bg4 Rb8 24.f4 Nd5 25.Qc1 c3 26.Kh1 Qc8 27.Bf3 Qc4 28.Be2 Qc5
29.Bf2 g5 30.fxg5 Bxg5 31.Qc2 Nf4 32.Bf3 Bxf3 33.Nxf3 Rxd1+ 34.Rxd1 Qxa3
35.Qe4 Qb3 36.Rg1 c2 37.Nxg5 hxg5 38.Be3 Qd3 39.Qxd3 Nxd3 40.h3 Rb1 41.Kh2
a4 42.g4 Rxg1 43.Kxg1 a3 44.Kf1 a2 45.Ke2 c1=N+ 46.Bxc1 Nxc1+ 47.Kd2 a1=Q
48.Kc2 Kg7 49.h4 gxh4 50.g5 h3 51.g6 Kxg6 52.Kd2 Qb2+ 53.Ke3 Kg5 54.Ke4
Qb4+ 55.Ke3 Qf4# 0-1

[Event "Computer chess game"]
[Site "LAZARUS"]
[Date "2004.07.19"]
[Round "2"]
[White "TSCP-172"]
[Black "Cilian-413"]
[Result "1/2-1/2"]
[ECO "C45e"]
[Variation "Scotch: Steinitz Variation"]
[TimeControl "21600"]

1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.Nxd4 Qh4 5.Nxc6 Qxe4+ 6.Be2 dxc6 7.O-O Be6
8.Re1 Be7 9.Nc3 Qf5 10.Bf3 Nf6 11.Ne2 Bc5 12.Ng3 Bxf2+ 13.Kxf2 Ng4+ 14.Kf1
Nxh2+ 15.Kf2 Ng4+ 16.Kf1 Nh2+ 17.Kf2 Ng4+ 18.Kf1 Nh2+ 19.Kf2 1/2-1/2

[Event "Computer chess game"]
[Site "MMCKEE_NT"]
[Date "2002.01.06"]
[Round "2"]
[White "TSCP-152"]
[Black "Fortress-162"]
[Result "1/2-1/2"]
[ECO "B56"]
[Variation "Sicilian: Open, 2...d6 5.Nc3 Nc6"]
[TimeControl "14400"]

1.e4 c5 2.Nf3 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 d6 6.Bb5 Bd7 7.Bg5 e5 8.Nf3
h6 9.Bxf6 Qxf6 10.O-O O-O-O 11.Nd5 Qe6 12.c4 f5 13.exf5 Qxf5 14.Qa4 Kb8
15.Rad1 g5 16.Qa3 Rh7 17.Rfe1 Rf7 18.Ne3 Qf4 19.Nd5 Qf5 20.Ne3 Qf4 21.Nd5
Qf5 22.Ne3 1/2-1/2

[Event "Computer chess game"]
[Site "MMCKEE_NT"]
[Date "2002.01.06"]
[Round "1"]
[White "Fortress-162"]
[Black "TSCP-172"]
[Result "0-1"]
[ECO "D15v"]
[Variation "Slav: Geller Gambit, 6.e5"]
[TimeControl "14400"]

1.d4 d5 2.c4 dxc4 3.Nf3 Nf6 4.Nc3 c6 5.e4 b5 6.e5 Nd5 7.Bd2 e6 8.Be2 Bb4
9.Ne4 Bxd2+ 10.Qxd2 O-O 11.Nd6 Na6 12.Ng5 f6 13.Nge4 Bd7 14.exf6 Nxf6 15.
O-O Nc7 16.Nxf6+ Qxf6 17.Ne4 Qe7 18.Bf3 Nd5 19.a4 a6 20.Nc5 Qd6 21.Rfe1
Ra7 22.Bg4 Re8 23.Bh5 Re7 24.Re5 Ra8 25.Bg4 Ree8 26.Rae1 Nc7 27.Qe3 bxa4
28.Nxa4 Rab8 29.Qg3 Rb4 30.Nc5 Rxb2 31.R5e2 Rxe2 32.Bxe2 Qxg3 33.fxg3 Bc8
34.Bxc4 Kf7 35.Kf2 Ke7 36.Ke3 Rd8 37.Ra1 Nb5 38.Bxb5 cxb5 39.Nxa6 Bb7 40.
Nc7 Bc6 41.Ra5 Rb8 42.Na6 Rb7 43.Nc5 Rc7 44.Nd3 Be8 45.Ne5 b4 46.Ra8 Rb7
47.Ra1 b3 48.Kd3 Kd6 49.Kc3 Bg6 50.Ra8 Bc2 51.Nc4+ Ke7 52.Rg8 Kf6 53.Rc8
Bd1 54.Rf8+ Kg5 55.Nb2 Rc7+ 56.Kb4 Bc2 57.Rf4 Ra7 58.Rf1 Rb7+ 59.Kc3 Be4
60.Na4 Bd5 61.Re1 Kf5 62.Nc5 Rb5 63.Re2 Kg5 64.Nd7 Kg4 65.Rf2 h5 66.Ne5+
Kg5 67.Rf8 h4 68.gxh4+ Kxh4 69.g3+ Kh3 70.Rh8+ Kg2 71.Nd3 Rb7 72.Kb2 Kf3
73.Rh5 g6 74.Rg5 Ke3 75.Nf4 Kxd4 76.Rxg6 e5 77.Nxd5 Kxd5 78.h4 e4 79.Rg8
e3 80.Re8 Kd4 81.g4 Rb5 82.g5 Re5 83.Rd8+ Kc4 84.g6 Re7 85.Rc8+ Kb4 86.
Rb8+ Kc5 87.Rc8+ Kd4 88.Rd8+ Ke4 89.g7 Rxg7 90.Kxb3 e2 91.Re8+ Kf3 92.Rf8+
Ke3 93.Re8+ Kd2 94.Rd8+ Ke1 95.Rf8 Rg2 96.Kb4 Rf2 97.Re8 Kf1 98.Kc5 e1=Q
99.Rxe1+ Kxe1 100.Kd5 Rh2 101.Kc5 Rxh4 102.Kd5 Kd2 103.Ke5 Kd3 104.Kd5
Rh5+ 105.Ke6 Ke4 106.Kd6 Kd4 107.Ke6 Re5+ 108.Kd6 Re2 109.Kd7 Kc5 110.Kc7
Rd2 111.Kb7 Rd7+ 112.Kb8 Kb6 113.Kc8 Rd2 114.Kb8 Rd8# 0-1

[Event "Computer chess game"]
[Site "MMCKEE_NT"]
[Date "2002.03.03"]
[Round "2"]
[White "TSCP-172"]
[Black "FRANWB-081"]
[Result "1-0"]
[ECO "C45k"]
[Variation "Scotch: Mieses, 8.c4"]
[TimeControl "14400"]

1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.Nxd4 Nf6 5.Nxc6 bxc6 6.e5 Qe7 7.Qe2 Nd5 8.c4
Nb6 9.Nc3 d6 10.exd6 cxd6 11.Bg5 Qe6 12.O-O-O Qxe2 13.Bxe2 Be6 14.c5 dxc5
15.Bf3 Bd5 16.Rhe1+ Kd7 17.Nxd5 cxd5 18.Bxd5 Nxd5 19.Rxd5+ Kc6 20.Red1 Re8
21.Be3 a6 22.Rd7 f6 23.Ra7 Kb5 24.Rdd7 Rb8 25.Bf4 Rc8 26.Rdb7+ Kc6 27.Rb3
c4 28.Rxa6+ Kd5 29.Rb5+ Ke4 30.Be3 Kd3 31.Kd1 Rd8 32.a4 c3 33.bxc3 Kxc3+
34.Ke2 Rc8 35.Ra7 Bd6 36.Rd7 Rhd8 37.Rxg7 Rg8 38.Rd7 Rcd8 39.Rxh7 Rge8 40.
h4 Re4 41.a5 Ra4 42.h5 Kc4 43.Rhb7 Ra3 44.h6 Re8 45.h7 Rh8 46.R5b6 Be5 47.
a6 Rxe3+ 48.fxe3 Bb8 49.Rxb8 Rxh7 50.R6b7 Rh2 51.e4 Kd4 52.Kf3 Rh1 53.Rd7+
Kc5 54.a7 Rf1+ 55.Ke2 Ra1 56.a8=Q Rxa8 57.Rd5+ Kc6 58.Rxa8 f5 59.exf5 Kxd5
60.Re8 Kd4 61.f6 Kc5 62.f7 Kd5 63.g4 Kd6 64.f8=Q+ Kc7 65.Re6 Kb7 66.Qf7+
Ka8 67.Re8# 1-0

[Event "Computer chess game"]
[Site "MMCKEE_NT"]
[Date "2003.02.24"]
[Round "1"]
[White "Gerbil-01"]
[Black "TSCP-152"]
[Result "0-1"]
[ECO "D31c"]
[Variation "QGD: 3.Nc3 Bb4"]
[TimeControl "14400"]

1.c4 e6 2.Nc3 d5 3.d4 Bb4 4.cxd5 Bxc3+ 5.bxc3 exd5 6.e3 Nc6 7.Nf3 Nge7 8.
Bd2 O-O 9.Bd3 Bf5 10.Bxf5 Nxf5 11.Qb3 Rb8 12.O-O-O b5 13.a4 bxa4 14.Qxa4
Qd6 15.h3 a5 16.c4 dxc4 17.g4 Qd5 18.gxf5 Qxf5 19.Qa2 Qxf3 20.Qxc4 Rb6 21.
d5 Rfb8 22.Qa2 Qf5 23.e4 Qxe4 24.Bc3 Rb1+ 25.Kd2 Qf4+ 26.Kc2 Qxf2+ 27.Bd2
Nb4+ 28.Kxb1 Qf5+ 29.Ka1 Qf6+ 30.Bc3 Qxc3+ 31.Qb2 Nc2+ 32.Kb1 Qxb2# 0-1

[Event "Computer chess game"]
[Site "MMCKEE_NT"]
[Date "2003.02.24"]
[Round "2"]
[White "TSCP-152"]
[Black "Gerbil-01"]
[Result "1/2-1/2"]
[ECO "C46"]
[Variation "Three Knights: Steinitz Variation"]
[TimeControl "14400"]

1.e4 e5 2.Nc3 Nc6 3.Nf3 g6 4.Bc4 Bg7 5.d3 Nf6 6.Ng5 O-O 7.O-O Na5 8.Nxf7
Rxf7 9.Bxf7+ Kxf7 10.f4 d6 11.fxe5 dxe5 12.Nd5 h6 13.Qe1 Nc6 14.Bxh6 Bxh6
15.Rxf6+ Kg7 16.Qg3 g5 17.h4 g4 18.Rxc6 bxc6 19.Qxe5+ Kg6 20.Ne7+ Kh7 21.
Rf1 Qe8 22.Qxc7 Be3+ 23.Kh1 Bd7 24.Nf5 Bc5 25.Qf4 Qh5 26.d4 Bxf5 27.dxc5
Bg6 28.Qc7+ Kh6 29.Qf4+ Kh7 30.Qc7+ Kh6 31.Qf4+ Kh7 1/2-1/2

[Event "Computer chess game"]
[Site "MMCKEE_NT"]
[Date "2003.02.25"]
[Round "1"]
[White "Gerbil-01"]
[Black "TSCP-172"]
[Result "0-1"]
[ECO "B00q"]
[Variation "Nimzowitsch Defence: 2.d4 d5"]
[TimeControl "14400"]

1.e4 Nc6 2.d4 d5 3.e5 Bf5 4.Nf3 e6 5.Nc3 Bb4 6.Bd2 Nge7 7.h3 O-O 8.g4 Be4
9.Nxe4 dxe4 10.Bxb4 Nxb4 11.Nd2 Qxd4 12.c3 Nd3+ 13.Bxd3 Qxd3 14.Qe2 Qxe2+
15.Kxe2 Ng6 16.Nxe4 Nxe5 17.f4 Ng6 18.Raf1 f5 19.Nc5 fxg4 20.Nxe6 Rae8 21.
f5 g3 22.Rhg1 Nh4 23.Kd3 Rxf5 24.Nxc7 Rd8+ 25.Ke4 Rxf1 26.Rxf1 g2 27.Rg1
Rf8 28.Rd1 g5 29.c4 Rf4+ 30.Ke3 Rf1 31.Rd8+ Kf7 32.Kd2 g1=Q 33.Rd7+ Kg6
34.Kc3 Qh2 35.b3 a5 36.a3 Rc1+ 37.Kd4 Qf2+ 38.Kd3 Qf3+ 39.Kd2 Qc3+ 40.Ke2
Re1+ 41.Kf2 Qe3# 0-1

[Event "Computer chess game"]
[Site "MMCKEE_NT"]
[Date "2003.02.25"]
[Round "2"]
[White "TSCP-172"]
[Black "Gerbil-01"]
[Result "1-0"]
[ECO "C45w"]
[Variation "Scotch: 4.Nxd4 Bc5 5.Be3 Qf6 6.c3 Nge7"]
[TimeControl "14400"]

1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.Nxd4 Bc5 5.Be3 Qf6 6.c3 Nge7 7.Be2 Bb6 8.O-O
d5 9.Nb5 a6 10.Bxb6 axb5 11.Bxc7 dxe4 12.Bxb5 Bf5 13.Bd6 O-O-O 14.Bxe7
Nxe7 15.Qb3 g5 16.Na3 Nd5 17.Bc4 Qe5 18.Nb5 Be6 19.Rad1 f5 20.Rfe1 Bf7 21.
Na7+ Kc7 22.Rd4 Rd7 23.g4 Rhd8 24.gxf5 Re7 25.Bxd5 Bxd5 26.Nb5+ Kc8 27.Qa4
Red7 28.c4 Bc6 29.Na7+ Kc7 30.Nxc6 bxc6 31.Qa7+ Kc8 32.Qa6+ Kc7 33.Rdxe4
Qxb2 34.Qa7+ Kc8 35.Qc5 Rc7 36.a4 Qf6 37.Qa5 Qb2 38.Re8 Qb7 39.f6 Qa7 40.
Qf5+ Kb7 41.Rb1+ Ka6 42.Rxd8 h6 43.Rd6 Qb7 44.Rxb7 Kxb7 45.Rd7 Kb6 46.Qb1+
Kc5 47.Rxc7 Kd6 48.Qb6 Ke5 49.f7 c5 50.f8=Q Ke4 51.Qbd6 g4 52.Qff4# 1-0


[Dan Honeycutt]

I think I'll try Simon next (if I am able to compile it in Mac OS X). Dan, do you have any estimate of how much weaker it is compared to TSCP?

Hi Tord:

Joshua Shriver did an OS X compile he might share, but I don't think you will have trouble. Platform dependent stuff is all in system.h

I don't any data on how it compares to TSCP. It's weaker than Gerbil and Faile which I think of as more or less comparable to TSCP.

Best
Dan H.

[Tord Romstad]

Hi Dan!

I think I'll try Simon next (if I am able to compile it in Mac OS X). Dan, do you have any estimate of how much weaker it is compared to TSCP?

Joshua Shriver did an OS X compile he might share, but I don't think you will have trouble. Platform dependent stuff is all in system.h

Yes, it compiled straight out of the box, with (literally!) thousands of warnings, but no errors. It seems to work fine. I played two quick test games against Glaurung with an Elo setting of 1700 (the result was 1-1, with one win for each side), and there were no problems at all.

All warnings were of this type:

Code: Select all

rand.cpp:288: warning: integer constant is too large for "long" type

I use the ULL suffix for 64-bit unsigned integers to avoid this warning. I am not sure whether this is standard C or just some gcc-specific thing. 64-bit ints seem to be poorly standardized in C.

Tord

[Dan Honeycutt]

Yes, it compiled straight out of the box, with (literally!) thousands of warnings, but no errors.

Hi Tord:

Looks like it gave a warning for every 64 bit number. rand.cpp (for the book) has 1024 random 64 bit numbers and simon has several array[64] of 64 bit numbers. The 64 bit data type definition looks like:

Code: Select all

#ifdef WIN32
  typedef unsigned __int64 U64;
#else
  typedef unsigned long long U64;
#endif


Maybe some compiler guru can say if that can be fixed to eliminate the warnings. But, at least the warnings seem to cause no harm.

Dan H.

[Uri Blass]

I see

66 TSCP-152 : 2116 117 59 70 21.4 % 2342 17.1 %
67 Beowulf-16b : 2114 143 70 50 20.0 % 2355 16.0 %
68 Gerbil-01 : 2087 75 30 247 22.7 % 2300 8.1 %

I do not believe that a single version of gerbil was weaker than tscp.

I believe that if engines do not score between 40-60% the rating can be misleading and the problem is not only statistical error.

When you do not care to give every player except the weaker players and the strongest players opponents that they score near than 50%your rating is simply misleading.

on the other hand if you do like Igor does in the infinite loop and give players to play almost only against players with similiar strength your rating is also misleading.

Your rating have difference that is too small when Igor's rating in the infinite loop has difference that is too high.

Uri

TSCP did well in my previous contest "Battle of the Crowns" also:

Code: Select all

    Program          Score     %    Av.Op.  Elo    +   -    Draws

  1 TSCP         :  30.0/ 36  83.3   2275   2555   83 155   22.2 %
  2 Monik        :  25.0/ 36  69.4   2289   2432   96 146   11.1 %
  3 Storm        :  21.5/ 36  59.7   2297   2365  107 112   19.4 %
  4 Noonian      :  21.0/ 36  58.3   2298   2356  109  78   44.4 %
  5 Ozwald       :  19.5/ 36  54.2   2300   2330  115 104   19.4 %
  6 Zephyr       :  19.5/ 36  54.2   2300   2330  115  97   25.0 %
  7 LarsenVB     :  19.5/ 36  54.2   2300   2330  115  97   25.0 %
  8 SnailSCP     :  10.0/ 36  27.8   2320   2154  129  93   22.2 %
  9 Raffaela     :   9.0/ 36  25.0   2322   2132  135  90   22.2 %
 10 Golem01      :   5.0/ 36  13.9   2335   2018  237  81    5.6 %


(the Elo rankings are only relatvie, of course)

I think that at long time control, TSCP is stronger than you think.

I think the opposite
Tscp is a weak program and like most of the weaker programs it is weaker at long time control because of worse branching factor.

Gerbil is using null move pruning and hash tables for better order of moves so Gerbil is stronger at long time control relative to tscp.

I remember that I played in the past some games between old movei
and tscp when I told movei to play 5 times faster.

It was with the first public version of movei(007a that is more than 400 elo weaker than movei of today).

I also told it always to change the first move(1.a3,1.a4,...)

I remember that in these conditions movei lost against tscp at fast time(1 minute against 5 minutes) but when I moved to slower time control movei beated tscp

I expect glaurung also to beat tscp at longer time control if tord tell Glaurung to play 200 times slower but not to do it at blitz.

Glaurung has a better branching factor than tscp and this is decisive at long time control even if tscp gets hardware that is 200 times faster.

[Uri Blass]

Dann,I can also add that long time control on the slow hardware that you used at that time is equivalent to blitz on fast hardware.

PII300 is probably more than 10 times slower than my A3000

Uri

[Dann Corbit]

Probably the stuff it is barking about is math on numbers that are too small.

For instance:

unsigned long long Two_to_the_thirty_third = 1 << 33;

is bad, because the one is undecorated and therefore too small to get a shift that big. Sometimes, a compiler will be smart enough to 'do what you wanted instead of what you said' and fix it for you.

Similarly,

unsigned long long bigprod = 1000000000 * 77;

is also wrong for the same sort of reason.