THE SSDF RATING LIST 2018-01-20

[Lars Sandin]

Code: Select all

THE SSDF RATING LIST 2018-01-20   %143840 games played by  364 computers							
                                           Rating   +	  -  Games   Won  Oppo
                                           ------  ---   --- -----   ---  ----
   1 Stockfish 8 MP x64 16GB 1800X 3,6 GHz   3436   52   -48   202   62%  3344
   2 Komodo 11.01 MP x64 16GB 1800X 3,6 GHz  3399   29   -27   854   77%  3192
   3 Deep Shredder 13 x64 16GB 1800X 3,6 GHz 3374   34   -31   560   72%  3209
   4 Komodo 9.1 MP x64 2GB Q6600 2,4 GHz     3353   24   -22  1273   77%  3142
   5 Komodo 11.01 MP x64 2GB Q6600 2,4 GHz   3340   49   -48   202   56%  3300
   6 Stockfish 6 MP x64 2GB Q6600 2,4 GHz    3333   25   -23  1016   72%  3166
   7 Deep Shredder 13 x64 2GB Q6600 2,4 GHz  3297   25   -24   844   63%  3203
   8 Komodo 7.0 MP x64 2GB Q6600 2,4 GHz     3271   26   -25   854   69%  3130
   9 Komodo 5.1 MP x64 2GB Q6600 2,4 GHz     3244   23   -22   998   65%  3134
  10 Deep Hiarcs 14 16GB 1800X 3,6 GHz       3216   32   -33   440   45%  3251
  11 Stockfish 3 MP x64 2GB Q6600 2,4 GHz    3204   19   -19  1340   62%  3116
  12 Deep Rybka 4 x64  2GB Q6600 2,4 GHz     3201   21   -20  1288   67%  3080
  13 Deep Rybka 3 x64  2GB Q6600 2,4 GHz     3193   22   -21  1371   75%  3002
  14 Deep Hiarcs 14 2GB Q6600 2,4 GHz        3188   19   -18  1410   61%  3107
  15 Chiron 3.01 MP x64 2GB Q6600 2,4 GHz    3180   28   -29   576   47%  3203
  16 Wasp 2.01 MP x64 16GB 1800X 3,6 GHz     3155   32   -34   446   39%  3229
  17 Naum 4.2 MP x64 2GB Q6600 2,4 GHz       3146   21   -21  1123   60%  3077
  18 Deep Junior Yokoh x64 2GB Q6600 2,4 GHz 3122   24   -24   827   46%  3152
  19 Naum 4 x64  2GB Q6600 2,4 GHz           3120   19   -19  1396   63%  3028
  20 Deep Junior 13.3 x64 2GB Q6600 2,4 GHz  3118   19   -19  1290   50%  3120
  21 Hiarcs 14 256MB Athlon 1200 MHz         3105   34   -33   440   62%  3020
  22 Spike 1.4 MP 2GB Q6600  2,4 GHz         3103   17   -16  1711   55%  3068
  23 Deep Shredder 12 x64 2GB Q6600 2,4 GHz  3103   18   -18  1507   62%  3015
  24 Hiarcs 13.1  2GB Q6600 2,4 GHz          3099   24   -23   868   56%  3054
  25 Deep Fritz 13 2GB Q6600 2,4 GHz         3097   24   -24   826   55%  3064
  26 Deep Hiarcs 13.2  2GB Q6600 2,4 GHz     3096   24   -24   832   55%  3064
  27 Deep Fritz 12 2GB Q6600 2,4 GHz         3090   20   -20  1200   55%  3054
  28 Deep Rybka 3  256MB Athlon 1200 MHz     3074   39   -37   332   58%  3019
  29 Deep Junior 12 x64 2GB  Q6600 2,4 GHz   3070   20   -20  1198   57%  3018
  30 Wasp 2.01 MP x64 2GB Q6600 2,4 GHz      3068   44   -52   320   20%  3299
  31 Zappa Mexico II x64  2GB Q6600 2,4 GHz  3062   21   -21  1026   54%  3035
  32 Deep Fritz 11  2GB Q6600 2,4 GHz        3059   18   -18  1504   62%  2975
  33 Naum 3.1 x64  2GB Q6600 2,4 GHz         3048   22   -22   986   49%  3052
  34 Crafty 25.0 MP x64 2GB Q6600 2,4 GHz    3023   25   -26   764   37%  3119
  35 Arasan 17.2 MP x64 2GB Q6600 2,4 GHz    2999   26   -26   686   46%  3028
  36 Naum 4  256MB Athlon 1200 MHz           2998   24   -24   836   40%  3069
  37 Fritz 13 256MB Athlon 1200 MHz          2996   45   -41   280   65%  2883
  38 Glaurung 2.2 x64 MP 2GB Q6600 2,4 GHz   2994   20   -19  1225   55%  2961
  39 Shredder 12 256MB A1200 MHz             2982   25   -27   760   35%  3090
  40 Wasp 2.01 256 MB Athlon 1200 MHz        2967   35   -34   417   60%  2897
  41 Revelation II Hiarcs 14.1               2923   47   -46   220   55%  2889
  42 CM King 3.5 x64 MP 2GB  Q6600 2,4 GHz   2863   24   -26   892   31%  3004
  43 Pro Deo 2.1 YAT 256MB Athlon 1200 MHz   2820   34   -33   440   59%  2758
  44 Revelation Hiarcs 13.3 XScale 500 MHz   2771   57   -52   177   66%  2659
  45 Revelation Shredder 12  XScale 500 MHz  2704   60   -58   140   56%  2664
  46 Revelation Rybka 2.2  XScale 500 MHz    2629   47   -44   240   62%  2546
  47 Revelation Deep Sjeng 3.0               2599   68   -76   100   37%  2691
  48 R30 v. 2.5                              2273   42   -38   343   69%  2136
  49 Millenium ChessGenius Pro M4 120 MHz    2184   92   -72   100   76%  1990
  50 Millenium ChessGenius ARM M4 48 MHz     2086   57   -53   169   62%  2001

 1 Stockfish 8 MP x64 16GB 1800X 3,6 GHz, 3436											
Komo11 1800X      21-19    Komo9.1 Q6600     26-14    Komo11 Q6600      26-16
Stockf6 Q6600   25,5-14,5  DShre13 Q6600     27-13    

 5 Komodo 11.01 MP x64 2GB Q6600 2,4 GHz, 3340											
Stock8 1800X      16-26    Komo11 1800X      18-22    DShre13 1800X   20,5-19,5
DHiar14 1800X     26-14    Wasp201 Q6600     32-8     

 16 Wasp 2.01 MP x64 16GB 1800X 3,6 GHz, 3155											
Komo11 1800X     7,5-32,5  DShre13 1800X      9-31    Komo9.1 Q6600      6-36
Stockf6 Q6600    5,5-34,5  Komo7.0 Q6600   12,5-27,5  DHiar14 1800X   19,5-20,5
DHiar14 Q6600     19-21    Chiro3 Q6600      17-23    Spike14 Q6600   25,5-14,5
Zap!Mx2 Q6600   28,5-15,5  Naum3.1 Q6600     26-14    

 30 Wasp 2.01 MP x64 2GB Q6600 2,4 GHz, 3068											
Komo11 1800X     7,5-72,5  DShre13 1800X    4,5-35,5  Komo11 Q6600       8-32
DShre13 Q6600    6,5-33,5  DHiar14 1800X    9,5-30,5  DHiar14 Q6600     14-26
Chiro3 Q6600    15,5-24,5  

On our latest ratinglist we have a new leader! It's the Stockfish 8-program 
by Tord Romstad, Marco Costalba and Joona Kiiski. As the program is released 
without an own book, we have used the free opening book from Fauzi Dabat,
called "Aggressive 3.2 by Fauzi.abk". Even though it's early on, this 
strong combination has achieved an impressive rating of 3436 after the first 
202 played games! This is 37 points above Komodo 11.01 on second place, but
we will of course need to play more games to be able to shrink the error bars.

Another early bird is the Komodo 11.01-program by Larry Kaufman and
Mark Lefler, which we introduced on our Ryzen-hardware last time around. 
We now have enough games to present an early rating for it on our Q6600-level.
With a rating of 3340 after the first 202 games, it still trails its older 
9.1 program by 13 points. This relationship will probably be readjusted for 
the next list as more games are played.

Since last time we have also included our formerly tested Wasp 2.01 by John
Stanback on both the Q6600 and the Ryzen 1800X hardware. On our Q6600 it has
reached a rating of 3068 after the first 320 games, and on the Ryzen hardware
it has achieved a rating of 3155 after 446 games. As we also have tested it 
on our 32-bit, singlecore A1200-level, we can now compare it on three 
different levels. Judging from our testing, it seems like the difference is 
approximately 100 rating points per hardware level for this program. 

Compared to our former ratinglist, Millenium ChessGenius and Deep Hiarcs 14
1800X gained 10 respectively 9 points. Wasp 2.01 A1200 lost 21 points, Deep
Shredder 13 1800X lost 11 and Komodo 11.01 1800X lost 8 points. 

The next ratinglist will be released as soon as we have enough games to be
able to present one or more of the manually tested chess computers which we
test at the moment. They missed this list by a small margin, but hopefully 
we will be able to present them in our next ratinglist which we hope to 
release in March/April.

Lars Sandin 

[Ovyron]

Are there any plans to test Houdini?

[Lars Sandin]

Are there any plans to test Houdini?

No, unluckily we have no such plans at the moment.

We would of course love to include Houdini in our list, especially as it seems to be one of the strongest programs out there; but if a programmer is silent and won't reply to any mails/questions - there are nothing we could do about it really.

Best regards

[Ovyron]

But the other rating lists test the Houdini chess engine without needing to contact the programmer.

[Lars Sandin]

So? Why should that help us in any sense? Do they share the same testing methods as us?

They are of course free to do whatever they like and what feels ok for them. Nothing is wrong when doing a non-paid hobby on your own spare-time and how they run their testing is not for me to tell.

[Ovyron]

Hmm, what questions do you have that Robert Houdart isn't answering? Perhaps if someone else answered them you'd test Houdini?

[ernest]

Hmm, what questions do you have that Robert Houdart isn't answering?

+1

[Werewolf]

How come the Genius 3 engine, running on a 120 MHz machine (at position 49) is weaker than the London 68030 running at 33 MHz??

A 68030 is ancient now.

[Lars Sandin]

Some key question, regarding our way of testing, that actually are quite simple and which should take less than a minute to answer to. He has every right to ignore answering some type of questions, and I then have to respect his sudden silence as a way of showing us that he isn’t interested in seeing his program in our list. It’s a pity really, but his own free choice.

[Lars Sandin]

Even though the 68030 (for instance) is old, one shouldn't underestimate the huge amount of adaptation and optimisation that went on "back in the days" for these kinds of hardware.

I would love to see a modern program in a old dedicated 8-bit machine like for instance the Mephisto Milano/Nigel Short; with all its inherent limitations - and see how it would perform compared to the "oldies". My guess it's no mundane task to adapt to the environment either way.

The London/Genius 68030, and especially the older Vancouver/Lyon/Portorose/Almeria 68030 costed some serious money. The newer machines are sold for a penny compared to the old ones and gives very good Lang-performance for the buck IMHO.