CCRL 40/4: Free single-CPU engines

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

User avatar
Kirill Kryukov
Posts: 518
Joined: Sun Mar 19, 2006 4:12 am
Full name: Kirill Kryukov

CCRL 40/4: Free single-CPU engines

Post by Kirill Kryukov »

Updated comparison of free sinlge-CPU engines in CCRL blitz:

Code: Select all

 Rank                  Engine                   ELO   +    -   Score  AvOp  Games
    1 Rybka 1.0 Beta 64-bit                    2945  +17  -16  74.3% -177.4  1428
    2 Fruit 2.3 5-men-egbb                     2874  +20  -20  66.8% -119.2   896
    3 Glaurung 2.0.1 64-bit                    2834  +15  -15  48.5%  +11.3  1550
    4 Spike 1.2 Turin                          2825   +6   -7  44.9%  +35.2  9624
    5 Naum 2.0 32-bit                          2797   +8   -8  48.9%   +2.7  5893
    6 Alaric 707                               2767  +12  -12  49.1%   +4.6  2451
      Glaurung 1.2.1 32-bit                    2757   +8   -8  41.9%  +57.5  5917
    7 Scorpio 2.0 (5-men-egbb)                 2754  +15  -15  49.5%   +4.9  1552
    8 Movei 00.8.438(10 10 10)                 2731  +17  -17  49.8%   +2.1  1210
    9 Pro Deo 1.2                              2721  +12  -12  42.9%  +53.6  2659
   10 Delfi 5.1                                2715  +11  -11  51.6%  -10.9  2894
11-12 Frenzee 3.0 64-bit                       2712  +15  -15  49.2%   +6.2  1540
11-12 Zappa 1.1 64-bit                         2712  +15  -15  46.8%  +21.5  1472
   13 Slow Chess Blitz WV2.1                   2711   +9   -9  41.7%  +61.5  5346
   14 Pharaon 3.5.1                            2701  +11  -11  38.6%  +84.2  3361
   15 WildCat 7                                2698  +10  -11  48.4%  +12.3  3421
   16 List 5.12                                2696  +11  -11  49.1%   +6.8  2990
   17 Colossus 2007c                           2688  +16  -16  47.6%  +16.8  1408
   18 SOS 5.1                                  2684  +12  -12  46.9%  +21.1  2569
   19 Pseudo 0.7c                              2677  +11  -11  50.6%   -7.6  3029
   20 Ruffian 1.0.5                            2676  +13  -13  49.4%   +3.4  2318
   21 Booot 4.13.1                             2671  +14  -14  51.5%  -10.3  1746
   22 Aristarch 4.50                           2668  +12  -12  45.8%  +32.0  2739
   23 Petir 4.39                               2665  +13  -13  46.7%  +25.0  2224
   24 Jonny 2.83 32-bit                        2652  +14  -14  46.2%  +29.6  1980
   25 The Baron 1.8.1                          2650  +16  -16  48.8%   +9.3  1462
   26 BugChess2 1.5.2                          2648  +17  -17  51.4%  -11.3  1248
   27 Crafty 21.5 PS 64-bit                    2645  +15  -15  45.5%  +33.1  1615
   28 Anaconda 2.0.1                           2642  +13  -13  51.7%  -10.7  1998
29-30 Fritz 5.32                               2640  +14  -14  53.1%  -24.4  1752
29-30 Smarthink 0.17a                          2640  +15  -15  50.8%   -5.9  1562
   31 Ktulu 4.2                                2634  +17  -16  52.4%  -17.3  1277
   32 Thinker 4.7a                             2633  +12  -12  50.3%   -5.8  2440
   33 Ufim 8.02                                2616  +12  -12  51.1%  -11.0  2652
34-36 AnMon 5.60                               2615  +14  -14  48.8%   +9.0  1924
34-36 Little Goliath Evolution 3.12 (TB On)    2615  +17  -17  50.2%   -1.7  1183
34-36 Trace 1.37a                              2615  +15  -15  47.5%  +19.0  1672
   37 Yace 0.99.87                             2612  +17  -17  50.3%   -2.6  1152
   38 E.T. Chess 18.11.05                      2610  +13  -13  54.2%  -36.1  2427
   39 Hamsters 0.5                             2591  +18  -18  49.5%   +1.9  1108
   40 Green Light Chess 3.00                   2575  +19  -19  46.6%  +22.1   959
   41 Tao 5.6                                  2567  +19  -19  45.6%  +30.4   952
42-43 Abrok 5.0                                2563  +18  -19  48.5%   +3.9  1052
42-43 Amyan 1.597                              2563  +14  -14  45.9%  +29.1  2056
   44 SpiderChess 070603                       2560  +15  -15  45.3%  +32.0  1593
   45 Arasan 10.0 64-bit                       2550  +20  -20  43.8%  +42.2   861
   46 Quark 2.35                               2521  +28  -28  49.0%   +8.9   459
   47 Snitch 1.6.2 32-bit                      2519  +20  -20  42.4%  +53.5   938
   48 Kiwi 0.6d                                2515  +18  -18  45.9%  +28.8  1131
   49 Nejmet 3.07                              2514  +16  -16  44.0%  +41.7  1381
   50 Comet B68                                2480  +19  -20  37.8%  +87.2   981
   51 Sjeng 12.13                              2479  +22  -22  35.4% +104.7   768
(My custom selection of well-tested engines).

News: Glaurung 2.0.1 64-bit is now 3-rd free single-CPU engine, after Rybka and Fruit!

I decided to keep both Glaurung 2.0.1 and 1.2.1 listed because they are very different from each other. I think it's safe to say that they are two different engines.

Rating list
Results and performance table
Score table
Likelihood of superiority table
Ponder hit table

Quark 2.35 is under testing - it has only few games now so it's rank is still likely to change.

Best,
Kirill
Tony Thomas

Re: CCRL 40/4: Free single-CPU engines

Post by Tony Thomas »

Glaurung is indeed showing a great improvement here as well. I am not even using the 64bit version, so I wonder what this puppy can do if I were to buy a better computer. Currently it is showing an improvement of over 100 points here. I do think that the previous Glaurung vesions were bit unlucky, still I love the results. Same results here, 4th strongest free engine after Rybka, fruit and Toga.
User avatar
Kirill Kryukov
Posts: 518
Joined: Sun Mar 19, 2006 4:12 am
Full name: Kirill Kryukov

Re: CCRL 40/4: Free single-CPU engines

Post by Kirill Kryukov »

I am considering to start listing Toga in my selection, because Toga and Fruit seem to have somewhat diverged over the years of independent development. I did not decide yet though.

Correlation of Fruit and Toga versions in CCRL 40/4. Early versions are almost identical, recent ones are more different. Evaluation is still very near, but the more important ponder hit stats show some hope. More data is needed to be sure.

Best,
Kirill