SSDF Rating List 2009-04-10

[Thoralf Karlsson]

THE SSDF RATING LIST 2009-04-10 119166 games played by 311 computers
Rating + - Games Won Oppo
------ --- --- ----- --- ----
1 Deep Rybka 3 x64 2GB Q6600 2,4 GHz 3224 31 -28 803 84% 2941
2 Naum 4 x64 2GB Q6600 2,4 GHz 3134 29 -27 714 74% 2955
3 Zappa Mexico II x64 2GB Q6600 2,4 GHz 3073 30 -29 568 63% 2976
3 Deep Rybka 3 256MB Athlon 1200 MHz 3073 44 -44 253 55% 3039
5 Deep Fritz 11 2GB Q6600 2,4 GHz 3071 25 -24 904 68% 2937
6 Naum 3.1 x64 2GB Q6600 2,4 GHz 3047 33 -32 467 60% 2977
7 Deep Shredder 11 x64 2GB Q6600 2,4 GHz 3043 29 -28 604 62% 2956
8 Deep Hiarcs 12 2GB Q6600 2,4 GHz 3038 26 -25 781 64% 2934
9 Hiarcs 11.2 MP 2GB Q6600 2,4 GHz 3011 29 -29 588 55% 2976
10 Naum 4 256MB Athlon 1200 MHz 3004 47 -47 223 50% 3004
11 Deep Junior 10.1 2GB Q6600 2,4 GHz 2980 29 -29 564 49% 2985
12 Rybka 2.3.1 Arena 256MB Athlon 1200 MHz 2923 23 -23 920 53% 2904
13 Fritz 11 256MB Athlon 1200 MHz 2913 29 -29 584 49% 2920
14 Deep Fritz 8 2GB Q6600 2,4 GHz 2910 27 -28 669 39% 2987
15 Deep Shredder 11 256MB Athlon 1200 2908 33 -33 450 46% 2936
16 Rybka 1.2 256MB Athlon 1200 MHz 2904 25 -24 893 71% 2748
17 Shredder 8 MP 2GB Q6600 2,4 GHz 2901 33 -34 441 43% 2950
18 Hiarcs 11.1 256MB Athlon 1200 MHz 2876 23 -23 901 50% 2873
19 Junior 10.1 256MB Athlon 1200 MHz 2856 21 -21 1092 51% 2853
20 Deep Junior 8 2GB Q6600 2,4 GHz 2851 33 -34 460 38% 2941
21 Fritz 10 256MB Athlon 1200 MHz 2847 36 -34 423 68% 2712
22 Zap!Chess Zanzibar 256MB Athlon 1200 MH 2843 22 -22 998 52% 2832
23 Junior 10 256MB Athlon 1200 MHz 2842 23 -22 1002 65% 2729
24 Hiarcs 10 HypMod 256MB Athlon 1200 MHz 2833 20 -19 1416 69% 2696
25 Fruit 2.2.1 256MB Athlon 1200 MHz 2829 19 -19 1385 62% 2746
26 Shredder 10 UCI 256MB Athlon 1200 MHz 2827 20 -20 1246 58% 2769
27 Spike 1.2 256MB Athlon 1200 MHz 2814 26 -26 714 57% 2763
28 Rybka 1.0 beta 128MB K6-2 450 MHz 2773 64 -69 115 38% 2857
29 Chess Tiger 2007 256MB Athlon 1200 MHz 2768 27 -27 671 49% 2778
30 Spike 1.1 256MB Athlon 1200 MHz 2760 31 -30 537 57% 2707
31 Zap!Chess 256MB Athlon 1200 MHz 2733 30 -29 562 53% 2709
32 Gandalf 6.0 256MB Athlon 1200 MHz 2730 24 -24 835 56% 2688
33 Chess Tiger 2004 256MB Athlon 1200 MHz 2723 21 -21 1088 52% 2706
34 Chessmaster 9000 256MB Athlon 1200 MHz 2716 36 -35 385 56% 2676
35 Pro Deo 1.1 256MB Athlon 1200 MHz 2710 24 -23 876 57% 2656
36 Deep Sjeng 1.5a 256MB Athlon 1200 MHz 2670 31 -31 493 52% 2659
37 CEBoard Fruit 2.3.1 XScale 400 400 MHz 2656 72 -65 112 65% 2549
38 Revelation Rybka 2.2 XScale 500 MHz 2634 54 -51 183 61% 2551
39 Ruffian 2.0.0 256MB Athlon 1200 MHz 2628 49 -49 205 46% 2657
40 Gromit 3.11.9 256MB Athlon 1200 MHz 2606 44 -46 246 43% 2658
41 Yace Paderborn 256MB Athlon 1200 MHz 2604 35 -35 389 48% 2618
42 Crafty 19.17 256MB Athlon 1200 MHz 2527 41 -44 304 30% 2677
43 Pocket Fritz 2 XScale 400 MHz 2505 52 -50 193 60% 2436
44 Resurrection Rybka 2.2 StrongARM 203 MH 2497 47 -46 224 55% 2461
45 Resurrection Fruit '05 StrongARM 203 MH 2395 67 -63 120 60% 2320
46 Hiarcs 9.5a/9.6 Palm TungstenE OMAP 126 2394 35 -35 400 45% 2429
47 CEBoard Crafty 2004 HP RX4240 400 MHz 2374 60 -61 135 49% 2384
48 R30 v. 2.5 2273 41 -38 343 69% 2134
49 Chess Genius 1.4 SX1 OMAP 310 120 MHz 2150 50 -48 210 60% 2080
50 Chess Tiger 14.9 Palm m515 16MB 42MHz 2102 69 -74 100 39% 2181




2 Naum 4 x64 2GB Q6600 2,4 GHz, 3134
DRybka3 Q6600 14-26 Zap!Mx2 Q6600 20.5-19.5 DRybka3 A1200 24-16
DFrit11 Q6600 22-18 Naum3.1 Q6600 16-9 DShre11 Q6600 26-14
DHiarcs Q6600 2-4 Hiar112 Q6600 26-16 DJun101 Q6600 30-10
Rybk231 A1200 36-9 Fritz11 A1200 33-7 DFritz8 Q6600 32.5-7.5
DShre11 A1200 34-6 Hiar111 A1200 37.5-4.5 Juni101 A1200 37-3
DJunio8 Q6600 33-7 Fritz10 A1200 7-0 Zap!Zan A1200 36.5-5.5
Fritz 9 A1200 36-5 Hiarcs9 A1200 23-1

10 Naum 4 256MB Athlon 1200 MHz, 3004
DRybka3 Q6600 7.5-32.5 DFrit11 Q6600 17-23 DHiarcs Q6600 17-23
Glaurung Q660 9-3 Fritz11 A1200 28.5-13.5 DJunio8 Q6600 25.5-14.5
CT 2007 A1200 7-2

37 CEBoard Fruit 2.3.1 XScale 400 400 MHz, 2656
Pock Fritz 3H 4-10 CM9000 A1200 3-6 PF3 Shredder 4.5-4.5
Revelat Rybka 13.5-6.5 Resurre Rybka 15.5-4.5 Hiarcs 9.5a 16.5-3.5
CEBoard Craft 15.5-4.5


Alexandar Naumovs latest chess program Naum 4 has been
tested on Q6600 2,4 GHz and on Athlon 1200 MHz. Naum 4 Q6600
has taken the second place of the list with a rating of
3134 after 714 tournament games. The rating increase compared
to Naum 3.1 is 87 points!

Naum 4 Athlon 1200 Mhz has received a rating of 3004, 69 points
behind Deep Rybka 3 on the same hardware.

We also have a new non-PC entrant. CEBoard Fruit 2.3.1 XScale 400
with a program written by Fabien Letouzey has a rating of 2656,
making it the strongest entrant of it´s kind!

Our rating program doesn´t seem to accept more than 311 entrants, so we
are forced to remove older programs when newer ones are to be tested. So
the total number of computers seen on top of the rating list will not
increase anymore, and the number of games played will not necessarily
increase either.

At the moment SSDF is testing Glaurung 2.2 Q6600.

Thoralf Karlsson

[Dr.Wael Deeb]

Thanks for the update....
A little note though:
I think that the rating of the dedicated units are way too overrated and must be removed from the list....
Just my opinion....
Dr.D

[pichy]

Thanks for the update....
A little note though:
I think that the rating of the dedicated units are way too overrated and must be removed from the list....
Just my opinion....
Dr.D

The dedicated units are a little bit overrated, but the engines rating are way much overrated by at least 100 rating points.

PS: I believe that if you take the dedicated units and match them versus humans, it will give us a more realistic rating to compare them with. The reason why the dedicated units are overrated is because they are being matched versus overrated engines.

[Mike S.]

(...)
I think that the rating of the dedicated units are way too overrated and must be removed from the list....

The dedicated units are a little bit overrated, but the engines rating are way much overrated by at least 100 rating points.

These statements surprise me. Overrated compared to what? There is no sufficient calibration. Also, note that the ratings of almost all dedicateds didn't change since many years, anymore. So, their rating is not at all influenced by the new engines and PCs. When the last rating level reduction was done, linear by -100 points for all, there was much common agreement among dedicateds fans, that they were now underrated(!). Many of them are rated club players, and knew club tournaments where such comps participated, and how they performed.

Therefore, web sources dedicated to these comps often give the "old" SSDF rating, from the time before this reduction. Example:

http://www.schach-computer.info/wiki/in ... -Elo-Liste

It would be a pity if the dedicated "historical " chess computers would disappear from the list. Instead, I suggest to think about double entries for the same chess program or -engine, where the only difference is the hardware. From such entries, maybe only the most important or most reliable (most games) needs to be kept.

To find these, you need to take a look at the complete list,

http://ssdf.bosjo.net/rlwww091.txt

P.S. I think it was a very good decision to include Glaurung 2.2, as (at least one) new representative for the strong freeware. I think a top-10 rank can be predicted, for the next list.

[Dr.Wael Deeb]

Thanks for the update....
A little note though:
I think that the rating of the dedicated units are way too overrated and must be removed from the list....
Just my opinion....
Dr.D

The dedicated units are a little bit overrated, but the engines rating are way much overrated by at least 100 rating points.

PS: I believe that if you take the dedicated units and match them versus humans, it will give us a more realistic rating to compare them with. The reason why the dedicated units are overrated is because they are being matched versus overrated engines.

I can asure you that in a 10 games match I can crush over 90% of all the dedicated units ever being made....and I play around 2100-2150 Elo....
Dr.D

[Mike S.]

You can crush a Tasc R30?! - If yes, then it is admirable, but it would make me think YOUR rating should be bigger.

A chess shop owner told me once, the CXG Sphinx Dominator (same as Galaxy) participated in a tournament in his chess club, won this tournament and performed (IIRC) above 1900. That comp was rated 1880 then, and is now rated 1780 only.

There must have been many more impressions like this, causing the common opinion mentioned.

Unfortunately, I do not have statistics from the Aegon tournaments, where many such comps participated against 2000+ humans (up to GMs). But I do not recall from the reports that anyone would have claimed that the comps were overrated.

I will not insist on it because I don't have enough good data to back it up, also different rating levels (in different pools/federations) could influence this matter.

[Dr.Wael Deeb]

You can crush a Tasc R30?! - If yes, then it is admirable, but it would make me think YOUR rating should be bigger.

A chess shop owner told me once, the CXG Sphinx Dominator (same as Galaxy) participated in a tournament in his chess club, won this tournament and performed (IIRC) above 1900. That comp was rated 1880 then, and is now rated 1780 only.

There must have been many more impressions like this, causing the common opinion mentioned.

Unfortunately, I do not have statistics from the Aegon tournaments, where many such comps participated against 2000+ humans (up to GMs). But I do not recall from the reports that anyone would have claimed that the comps were overrated.

I will not insist on it because I don't have enough good data to back it up, also different rating levels (in different pools/federations) could influence this matter.

With a good preparation for the match and clear head,I think that my chances are rather very good to excellent....

BTW,they say that playing the chess engines won't help you playing stronger against the humans....definitely wrong....
After playing 612 long time control games against the chess engines since the year 2002,I rised my own Elo about 350-400....all my friends and relatives were crushed like a bug daring to play me....the last one I played a month ago told me that he would call a russian friend of his to crush me....I am still waiting for the russian

Dr.D

[Dann Corbit]

(...)
I think that the rating of the dedicated units are way too overrated and must be removed from the list....

The dedicated units are a little bit overrated, but the engines rating are way much overrated by at least 100 rating points.

These statements surprise me. Overrated compared to what? There is no sufficient calibration. Also, note that the ratings of almost all dedicateds didn't change since many years, anymore. So, their rating is not at all influenced by the new engines and PCs. When the last rating level reduction was done, linear by -100 points for all, there was much common agreement among dedicateds fans, that they were now underrated(!). Many of them are rated club players, and knew club tournaments where such comps participated, and how they performed.

Therefore, web sources dedicated to these comps often give the "old" SSDF rating, from the time before this reduction. Example:

http://www.schach-computer.info/wiki/in ... -Elo-Liste

It would be a pity if the dedicated "historical " chess computers would disappear from the list. Instead, I suggest to think about double entries for the same chess program or -engine, where the only difference is the hardware. From such entries, maybe only the most important or most reliable (most games) needs to be kept.

To find these, you need to take a look at the complete list,

http://ssdf.bosjo.net/rlwww091.txt

P.S. I think it was a very good decision to include Glaurung 2.2, as (at least one) new representative for the strong freeware. I think a top-10 rank can be predicted, for the next list.

The rating of the dedicated machines is exactly correct, according to this list (the positions are formed using simple mathematics). If you try to bend it towards other scenarios (e.g. "What would happen if these machines played in a Fide tournament?") then of course none of the ratings will give a perfect answer. But we already knew that.

[Mike S.]

The rating of the dedicated machines is exactly correct, according to this list (the positions are formed using simple mathematics).

I understand what you mean, but I think it is possible that you are not aware of the SSDF history: At a certain point of time, ALL ratings have been reduced by 100 points. So, the rating difference I mentioned for the Sphinx Galaxy was NOT because his performance dropped while playing more games, it was because of this change of the total numbers level, so to speak.

See this page which lists the before/after ratings (1992/2000):

http://www.schachcomputer.at/eloliste.htm

(Only in very few cases, the difference is other than exactly 100, which must mean that the comp has played games after the change.)

It is possible that there had already been a similar reduction before, but I am not sure. If yes, than that was a smaller one.

I think the reason for that was each, that the ratings on the top of the list seemed "unrealistic" compared to FIDE Elo. It is true that in fact, this doesn't matter theoretically(!) at all as it are different pools. But in practice, if for example new cc. fans, or even journalists who may not be chess rating experts etc. see a couple of "Elo" rating lists, they will of course think a 2600 here is as strong as a 2600 there. This is a practical problem.

It is my interpretation that this is the reason why this ratings level reduction(s) were done, but I forgot the SSDF comments about it. Thoralf Karlsson could probably acknowledge it, or explain the other reasons if I am wrong.

OTOH, it would not be a big problem to display a big sign on the ratings website each, just saying

THESE RATINGS ARE NOT FIDE ELOS!

Anyway, my whole point was that the "old" ratings for the dedicated comps were considered more similar to federation ratings of humans with similar or same strengths.

[icander]

Code: Select all

THE SSDF RATING LIST 2009-04-10   %119166 games played by  311 computers
                                           Rating   +     -  Games   Won  Oppo
                                           ------  ---   --- -----   ---  ----
   1 Deep Rybka 3 x64  2GB Q6600 2,4 GHz     3224   31   -28   803   84%  2941
   2 Naum 4 x64  2GB Q6600 2,4 GHz           3134   29   -27   714   74%  2955
   3 Zappa Mexico II x64  2GB Q6600 2,4 GHz  3073   30   -29   568   63%  2976
   3 Deep Rybka 3  256MB Athlon 1200 MHz     3073   44   -44   253   55%  3039
   5 Deep Fritz 11  2GB Q6600 2,4 GHz        3071   25   -24   904   68%  2937
   6 Naum 3.1 x64  2GB Q6600 2,4 GHz         3047   33   -32   467   60%  2977
   7 Deep Shredder 11 x64 2GB Q6600  2,4 GHz 3043   29   -28   604   62%  2956
   8 Deep Hiarcs 12  2GB Q6600 2,4 GHz       3038   26   -25   781   64%  2934
   9 Hiarcs 11.2 MP  2GB  Q6600  2,4 GHz     3011   29   -29   588   55%  2976
  10 Naum 4  256MB Athlon 1200 MHz           3004   47   -47   223   50%  3004
  11 Deep Junior 10.1  2GB  Q6600  2,4 GHz   2980   29   -29   564   49%  2985
  12 Rybka 2.3.1 Arena 256MB Athlon 1200 MHz 2923   23   -23   920   53%  2904
  13 Fritz 11  256MB Athlon 1200 MHz         2913   29   -29   584   49%  2920
  14 Deep Fritz 8 2GB  Q6600  2,4 GHz        2910   27   -28   669   39%  2987
  15 Deep Shredder 11  256MB  Athlon 1200    2908   33   -33   450   46%  2936
  16 Rybka 1.2  256MB Athlon 1200 MHz        2904   25   -24   893   71%  2748
  17 Shredder 8 MP 2GB  Q6600  2,4 GHz       2901   33   -34   441   43%  2950
  18 Hiarcs 11.1  256MB  Athlon 1200 MHz     2876   23   -23   901   50%  2873
  19 Junior 10.1  256MB Athlon 1200 MHz      2856   21   -21  1092   51%  2853
  20 Deep Junior 8  2GB Q6600 2,4 GHz        2851   33   -34   460   38%  2941
  21 Fritz 10  256MB Athlon 1200 MHz         2847   36   -34   423   68%  2712
  22 Zap!Chess Zanzibar 256MB Athlon 1200 MH 2843   22   -22   998   52%  2832
  23 Junior 10  256MB Athlon 1200 MHz        2842   23   -22  1002   65%  2729
  24 Hiarcs 10 HypMod 256MB Athlon 1200 MHz  2833   20   -19  1416   69%  2696
  25 Fruit 2.2.1  256MB Athlon 1200 MHz      2829   19   -19  1385   62%  2746
  26 Shredder 10 UCI  256MB Athlon 1200 MHz  2827   20   -20  1246   58%  2769
  27 Spike 1.2  256MB Athlon 1200 MHz        2814   26   -26   714   57%  2763
  28 Rybka 1.0 beta 128MB K6-2 450 MHz       2773   64   -69   115   38%  2857
  29 Chess Tiger 2007  256MB Athlon 1200 MHz 2768   27   -27   671   49%  2778
  30 Spike 1.1  256MB Athlon 1200 MHz        2760   31   -30   537   57%  2707
  31 Zap!Chess  256MB Athlon 1200 MHz        2733   30   -29   562   53%  2709
  32 Gandalf 6.0  256MB  Athlon 1200 MHz     2730   24   -24   835   56%  2688
  33 Chess Tiger 2004  256MB Athlon 1200 MHz 2723   21   -21  1088   52%  2706
  34 Chessmaster 9000  256MB Athlon 1200 MHz 2716   36   -35   385   56%  2676
  35 Pro Deo 1.1  256MB Athlon 1200 MHz      2710   24   -23   876   57%  2656
  36 Deep Sjeng 1.5a  256MB Athlon 1200 MHz  2670   31   -31   493   52%  2659
  37 CEBoard Fruit 2.3.1 XScale 400  400 MHz 2656   72   -65   112   65%  2549
  38 Revelation Rybka 2.2  XScale 500 MHz    2634   54   -51   183   61%  2551
  39 Ruffian 2.0.0  256MB  Athlon 1200 MHz   2628   49   -49   205   46%  2657
  40 Gromit 3.11.9  256MB Athlon 1200 MHz    2606   44   -46   246   43%  2658
  41 Yace Paderborn  256MB Athlon 1200 MHz   2604   35   -35   389   48%  2618
  42 Crafty 19.17  256MB Athlon 1200 MHz     2527   41   -44   304   30%  2677
  43 Pocket Fritz 2 XScale 400 MHz           2505   52   -50   193   60%  2436
  44 Resurrection Rybka 2.2 StrongARM 203 MH 2497   47   -46   224   55%  2461
  45 Resurrection Fruit '05 StrongARM 203 MH 2395   67   -63   120   60%  2320
  46 Hiarcs 9.5a/9.6 Palm TungstenE OMAP 126 2394   35   -35   400   45%  2429
  47 CEBoard Crafty 2004 HP RX4240  400 MHz  2374   60   -61   135   49%  2384
  48 R30 v. 2.5                              2273   41   -38   343   69%  2134
  49 Chess Genius 1.4 SX1  OMAP 310 120 MHz  2150   50   -48   210   60%  2080
  50 Chess Tiger 14.9 Palm m515 16MB 42MHz   2102   69   -74   100   39%  2181