CCRL 40/4 - Updated Free Single-CPU comparison

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

User avatar
Kirill Kryukov
Posts: 519
Joined: Sun Mar 19, 2006 4:12 am
Full name: Kirill Kryukov

CCRL 40/4 - Updated Free Single-CPU comparison

Post by Kirill Kryukov »

Updated comparison of free single-CPU engines in CCRL 40/4 time control:

Code: Select all

 Rank                 Engine                   ELO   +    -   Score  AvOp  Games
    1 Rybka 1.0 Beta 64-bit                   2945  +16  -16  74.9% -182.0  1620
    2 Fruit 2.3                               2883  +18  -18  68.6% -132.5  1087
    3 Glaurung 2.0.1 64-bit                   2838   +9   -9  47.2%  +17.8  5243
    4 Spike 1.2 Turin                         2826   +6   -6  43.9%  +41.6 15022
    5 Naum 2.0 32-bit                         2795   +8   -8  48.9%   +3.0  6092
    6 Bright 0.2c                             2786  +14  -14  45.8%  +30.7  1803
    7 Alaric 707                              2760  +10  -10  44.5%  +38.3  3963
    8 Scorpio 2.0 32-bit                      2747  +10  -10  50.7%   -3.8  3179
    9 Movei 00.8.438(10 10 10)                2738  +13  -13  48.9%   +7.8  2127
   10 E.T. Chess 13.01.08                     2731  +14  -14  48.3%  +13.3  1756
   11 Alfil 8.1.1 Optimized                   2726  +17  -17  50.3%   -1.5  1152
   12 Pro Deo 1.6(2007-11-14)                 2717  +20  -20  48.0%  +15.7   864
13-14 Colossus 2007d                          2714  +13  -13  45.8%  +29.9  2048
13-14 Frenzee 3.0 64-bit                      2714  +14  -14  49.7%   +2.0  1828
   15 Slow Chess Blitz WV2.1                  2712   +8   -8  41.5%  +61.6  6561
16-17 Delfi 5.2                               2711  +11  -11  41.4%  +61.7  3164
16-17 Zappa 1.1 64-bit                        2711  +14  -14  47.2%  +19.2  1728
   18 WildCat 7                               2702   +9   -9  46.7%  +22.8  4599
   19 Pharaon 3.5.1                           2701  +11  -10  39.2%  +79.5  3642
   20 List 5.12                               2693  +11  -11  48.5%  +11.2  3214
   21 SOS 5.1                                 2684  +11  -12  46.7%  +23.1  2857
   22 Ruffian 1.0.5                           2680  +12  -12  49.2%   +4.7  2606
   23 Pseudo 0.7c                             2677  +11  -11  50.0%   -2.8  3317
   24 Booot 4.13.1                            2669  +13  -13  51.0%   -6.7  2096
   25 Aristarch 4.50                          2668  +11  -11  45.7%  +32.6  2962
   26 Hamsters 0.6                            2662  +14  -14  50.5%   -3.7  1771
   27 Petir 4.999999                          2659  +14  -14  48.4%  +10.7  1754
   28 Jonny 2.83 32-bit                       2652  +13  -13  45.9%  +30.9  2172
   29 The Baron 1.8.1                         2648  +14  -14  48.6%   +8.6  1779
   30 Crafty 21.5 PS 64-bit                   2644  +14  -14  45.7%  +31.5  1743
31-33 Anaconda 2.0.1                          2641  +13  -13  51.1%   -7.2  2189
31-33 BugChess2 1.5.2                         2641  +15  -15  50.6%   -6.1  1624
31-33 Fritz 5.32                              2641  +13  -13  53.1%  -24.0  2005
34-35 SmarThink 0.17a                         2634  +14  -14  50.9%   -6.3  1754
34-35 Thinker 4.7a                            2634  +12  -12  51.0%  -10.6  2631
   36 Ktulu 4.2                               2633  +15  -15  53.1%  -21.7  1469
   37 Ufim 8.02                               2625  +11  -11  51.8%  -15.7  3187
   38 Trace 1.37a                             2614  +14  -14  48.3%  +12.3  2000
39-40 AnMon 5.60                              2612  +13  -13  49.3%   +6.0  2148
39-40 Yace 0.99.87                            2612  +15  -15  51.4%  -10.8  1472
   41 Little Goliath Evolution 3.12           2609  +12  -12  52.9%  -22.9  2557
   42 Green Light Chess 3.00                  2575  +16  -16  48.1%  +12.0  1310
   43 Tao 5.6                                 2567  +17  -17  47.6%  +16.6  1271
   44 Abrok 5.0                               2564  +16  -16  48.7%   +5.0  1371
   45 Amyan 1.597                             2561  +13  -13  46.6%  +23.8  2344
   46 Twisted Logic 20080404x                 2560  +20  -20  49.6%   +2.4   928
   47 Arasan 10.0 64-bit                      2558  +18  -18  47.7%  +15.9  1117
   48 LambChop 10.99                          2554  +23  -23  52.7%  -20.5   639
   49 SpiderChess 070603                      2553  +12  -13  47.6%  +16.5  2311
   50 Matacz 1.3 HT74                         2539  +19  -19  46.4%  +25.3   926
   51 Pepito 1.59 64-bit                      2538  +18  -17  46.6%  +22.3  1119
   52 Kiwi 0.6d                               2519  +16  -16  46.2%  +26.5  1419
   53 Nejmet 3.07                             2518  +15  -15  44.9%  +35.5  1636
   54 Snitch 1.6.2 32-bit                     2515  +17  -17  43.6%  +45.4  1194
   55 Quark 2.35                              2514  +17  -17  43.4%  +47.9  1258
   56 Dragon 4.6                              2508  +15  -15  42.7%  +56.3  1614
57-58 Der Bringer 1.9                         2477  +16  -16  46.7%  +25.2  1367
57-58 Sjeng 12.13                             2477  +19  -19  37.3%  +90.5   992
   59 Comet B68                               2475  +17  -17  40.1%  +70.5  1366
   60 Phalanx XXII Reborn JA                  2434  +25  -25  36.3%  +97.8   576
Rating list and crosstable

Most recent additions were Matacz 1.3 HT74, Der Bringer 1.9, Phalanx XXII Reborn JA and LambChop 10.99.

I realize that many engines in my selection already have newer version, however resources available to me do not allow me to follow the busy update schedule of all engines. I (or others) will test those updated versions in due time, in the meantime I am looking into older strong engines (2005 and before).

This is not any "official" list, but my personal selection, where I include well-tested original free single-CPU engines.

Best,
Kirill
Edsel Apostol
Posts: 803
Joined: Mon Jul 17, 2006 5:53 am
Full name: Edsel Apostol

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Edsel Apostol »

Thanks Kirill, its nice to see my old release version gets into the top 50.
User avatar
Kirill Kryukov
Posts: 519
Joined: Sun Mar 19, 2006 4:12 am
Full name: Kirill Kryukov

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Kirill Kryukov »

Edsel Apostol wrote:Thanks Kirill, its nice to see my old release version gets into the top 50.
Hi Edsel! I suspect there are engines that may challenge its position in top 50. Though your latest version does not need to worry. I'm looking forward to test it, but my queue needs some clearing up before that. Best to you! Kirill
User avatar
Mike S.
Posts: 1480
Joined: Thu Mar 09, 2006 5:33 am

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Mike S. »

From the newer engines, Bright 0.3a is stronger than 0.2c. Thinker 5.x has joined the group in the range of Spike 1.2, Glaurung 2.1 etc. Also, I think Delfi 5.4 has made big progress. I think we will soon se a group with Glaurung,Bright,Thinker,Spike,Delfi and Twisted Logic within a narrow margin.

On the top, we can expect a free Rybka 2.2n2 soon, which will be released as a Rybka demo. The 64 bit version is ~80 Elo stronger and the 32 bit version ~100 Elo stronger than Rybka 1.0 beta. :mrgreen:
Regards, Mike
Tony Thomas

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Tony Thomas »

Mike S. wrote:From the newer engines, Bright 0.3a is stronger than 0.2c. Thinker 5.x has joined the group in the range of Spike 1.2, Glaurung 2.1 etc. Also, I think Delfi 5.4 has made big progress. I think we will soon se a group with Glaurung,Bright,Thinker,Spike,Delfi and Twisted Logic within a narrow margin.

On the top, we can expect a free Rybka 2.2n2 soon, which will be released as a Rybka demo. The 64 bit version is ~80 Elo stronger and the 32 bit version ~100 Elo stronger than Rybka 1.0 beta. :mrgreen:
Is there a reason why they decided to release 2.2n2 instead of 2.3.2? Is there any significant strength difference between the two? I do not own the commercial versions of Rybka, so I am kind of lost here..

Code: Select all

 1 RomiChessP3k RG                   3471  113  107    31   63%  3360   23% 
   2 RomiChessP3k WB                   3387  107  113    30   38%  3471   23% 
   3 Hiarcs12SPUCI                     3210   54   49   278   88%  2825   12% 
   4 Naum 3                            3146   48   45   204   77%  2927   25% 
   5 Hiarcs11.2SPUCI                   3133   51   49   186   74%  2934   16% 
   6 HiarcsX54UCI                      3086   46   44   208   71%  2918   18% 
   7 Naum 3.1                          3085   61   58   109   67%  2946   27% 
   8 Thinker 5.1b                      3080   83   77    71   70%  2914   15% 
   9 Rybka v1.0 Beta.w32               3080   36   35   322   70%  2926   25% 
Ignoring the random results of Romichess we can see that Rybka will probably end up in the top spot once again.
Michael Sherwin
Posts: 3196
Joined: Fri May 26, 2006 3:00 am
Location: WY, USA
Full name: Michael Sherwin

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Michael Sherwin »

Okay Tony, I will ignore those numbers! 8-) No problem, I am not curious in the slightest. And I am sure that no one else is either.

I just figure that you gave Romi such a high rating, because, you are "on the outs", with Goat! :lol:
If you are on a sidewalk and the covid goes beep beep
Just step aside or you might have a bit of heat
Covid covid runs through the town all day
Can the people ever change their ways
Sherwin the covid's after you
Sherwin if it catches you you're through
User avatar
Mike S.
Posts: 1480
Joined: Thu Mar 09, 2006 5:33 am

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Mike S. »

Tony Thomas wrote: Is there a reason why they decided to release 2.2n2 instead of 2.3.2?
It makes sense to me, in respect of the Rybka 2.x buyers. I bought it not before version 2.3.2a was released, on CD (the world championship version) in August 2007.

I think that 2.2, and most probably 2.2n2, is better in the late endgame (maybe except some zugzwang positions), than 2.3.2a which has some quirks in this area.

Nevertheless, I think I will stick to 2.2 for that purpose because I read, 2.2n2 doesn't display the tablebase hits. The improvements from 2.2 to 2.2n2 mostly were for mp effectiveness, but the demo will be the single version.

The free demo will also include the RybkaII.ctg book, which I think is quite remarkable. This will certainly be one of the very best free CTG books.

(I am referring to announcements, only. It's not yet up.)
Regards, Mike
User avatar
Kirill Kryukov
Posts: 519
Joined: Sun Mar 19, 2006 4:12 am
Full name: Kirill Kryukov

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Kirill Kryukov »

Mike S. wrote:From the newer engines, Bright 0.3a is stronger than 0.2c. Thinker 5.x has joined the group in the range of Spike 1.2, Glaurung 2.1 etc. Also, I think Delfi 5.4 has made big progress. I think we will soon se a group with Glaurung,Bright,Thinker,Spike,Delfi and Twisted Logic within a narrow margin.

On the top, we can expect a free Rybka 2.2n2 soon, which will be released as a Rybka demo. The 64 bit version is ~80 Elo stronger and the 32 bit version ~100 Elo stronger than Rybka 1.0 beta. :mrgreen:
All these engines are on my testing queue, and Rybka 2.2n2 will be added there as soon as it is released as free engine. :-)

I want to find out reliable top-40, top-50 and top-60 lists of free single-CPU engines, so I now have many older engines in my testing plan.
Tony Thomas

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Tony Thomas »

Michael Sherwin wrote:Okay Tony, I will ignore those numbers! 8-) No problem, I am not curious in the slightest. And I am sure that no one else is either.

I just figure that you gave Romi such a high rating, because, you are "on the outs", with Goat! :lol:
It was part of book testing for Romi for ACCAWCRCC where we did not inhale vigorously in goats view. The first one used the book from Robindu and the second one the book from Wael..When I merged the pgn, I didnt consider the fact that these Romi version (renamed Proto3k) didnt play against other engines. I never had thought of this way to make Romi the number one engine, so I thought it was hilarious and didnt recalculate the ratings with a ratings 40 command or something similar.
Tony Thomas

Re: CCRL 40/4 - Updated Free Single-CPU comparison

Post by Tony Thomas »

Mike S. wrote:
Tony Thomas wrote: Is there a reason why they decided to release 2.2n2 instead of 2.3.2?
It makes sense to me, in respect of the Rybka 2.x buyers. I bought it not before version 2.3.2a was released, on CD (the world championship version) in August 2007.

I think that 2.2, and most probably 2.2n2, is better in the late endgame (maybe except some zugzwang positions), than 2.3.2a which has some quirks in this area.

Nevertheless, I think I will stick to 2.2 for that purpose because I read, 2.2n2 doesn't display the tablebase hits. The improvements from 2.2 to 2.2n2 mostly were for mp effectiveness, but the demo will be the single version.

The free demo will also include the RybkaII.ctg book, which I think is quite remarkable. This will certainly be one of the very best free CTG books.

(I am referring to announcements, only. It's not yet up.)
Mike, thanks for your detailed reply. If I can have the free Rybka 2.2n then why would a computer chess addict like me buy Rybka 3 other than for the aquarium GUI? May be I should save those 60 dollars and use it to fill up my gas tank :wink: I am reasonably satisfied with Arena for testing engines, but it does have few unexplainable glitches.. Sometimes I start a tournament, and when I come back few hours later, I will see that all the games are finished, but none of the engines made any moves. The annotation on the pgn says human vs human, and I know for sure I didnt add any humans to the tournament. I really have no use for the .ctg book unless I buy Aquarium which I heard supports .ctg books. As I have stated many times, I wont buy/Use any Chessbase products unless they start supporting wb protocol the same way they support UCI.