Has anyone ever heard of an engine called Smash?

[Dr.Wael Deeb]

That might not mean a whole lot, as you are only a single opponent, and be especially experienced in dealing with computers.

I am not sure what all the numbers in your rating list mean, but if you group Smash with the other engines mentioned there, I think something must be wrong with the way you installed it. It is really incomparably stronger than most of those. The current version of micro-Max scores nearly 50% against it. And I have tested uMax against many of the engines on your list, during its development, and at the end it totally crushed most of them, to the extent that I had to replace them from the standard gauntlett, because playing them was basically a waste of time (contributing more to the noise than to the rating).

The numbers are simply the rating in Elo and the number of points gained from the number of the played games,I calculate the Elo manualy using a tool called ELO CALC....
i am sure that it's installed correctly and plays with maximum strength,it's weak and it have played head to head with these engines at long time controls,it plays at the level of Nero and Soldat,run a match with these two engines and see for yourself....

You are totaly misleaded by these Elo calculating programs

I don't know anything about your program regarding the strength as it's still waiting on my waiting list for testing

[Dr.Wael Deeb]

The latest smash version plays around 1800 Elo and I can confirm this fact from the games I've played against it,I can beat it easily in a 10 games match




Sorry Dr. Weeb

But I find this very hard to believe, I'm rated 2032 fide and played two serious games with Smash and I lost both games 40moves 1hour.
Also I'm trying to rate it against several chessmaster 10 personalities and after 6 games it has a rating of 2180, I don't know of any MP version I'm using the single version, but it is by no means 1800elo.

Hi,
No need to be sorry,it's just that the programs at this level of playing are poorly tested and people don't have enough information about them....

Please take a look here:

http://www.cli.di.unipi.it/~sambati/prog/smash.html

The latest version is Smash 1.0.3 and it's an MP program,but still,it's playing around 1800 Elo,download and try it,I'll be more than happy to hear your opinion

[hgm]

Well, from 3 points out of 68 games it is totally impossible to determine a meaningful ELo rating. You might as well play zero games and throw dice. Whatever you get, is all statistical error.

No wonder there is no relationship whatsoever between the strength of the engine and the rating you give. Most of them have only 3 points out of a very large number of games. That means that most of the games must have been played against opponents so much stronger, that the outcome gives zero information.

If you want proof: accordng to yur ratings, Smash and TSCP are practically equally strong, and a 10-game match between them will end close to 5-5. But I can predict you that it will be more like 9-1 for Smash, if TSCP is lucky.

[Dr.Wael Deeb]

I dont have as many games as CCRL but it is rated way lower...Ranked 158, the huge variation in rating could be due to Elostat messing up the rating of weaker engines.

Code: Select all

148 Clueless 1.3b                  : 1928   78  76    77    64.3 %   1826   11.7 %
149 BigLion 2.23w                  : 1924   71  69    94    64.4 %   1821    9.6 %
150 Laurifer 1.0                   : 1921   77  75    78    63.5 %   1825   11.5 %
151 BBChess 0.99b                  : 1920   75  74    82    62.2 %   1834    9.8 %
152 IACH 0.56E                     : 1919   72  71    84    61.9 %   1834   14.3 %
153 Ssechess 2.05DC                : 1918   77  76    78    62.8 %   1827   10.3 %
154 Alf 1.06                       : 1914   73  72    83    60.2 %   1842   12.0 %
155 Gerbil 02                      : 1912   74  72    80    62.5 %   1823   15.0 %
156 ChessAlex 1.4 r 3              : 1902   81  79    70    59.3 %   1837   10.0 %
157 FireFly 2.13                   : 1894   73  72    78    59.6 %   1827   16.7 %
158 Smash 1.03                     : 1891   69  68    83    57.8 %   1836   19.3 %
159 Gibbon 2.01b                   : 1889   75  74    73    56.2 %   1846   16.4 %
160 Feuerstein 0.4.5 beta 1        : 1872   74  73    73    55.5 %   1834   17.8 %
161 Warrior 1.0.3                  : 1869   76  76    71    55.6 %   1829   15.5 %
162 Mustang 3.0                    : 1863   75  75    70    55.0 %   1828   18.6 %
163 Marvin 1.3.0                   : 1848   73  73    79    53.2 %   1826   12.7 %
164 FreeChess 2.16e                : 1832   72  72    77    50.6 %   1827   15.6 %
165 Enigma 1.1.4                   : 1805   75  75    77    46.8 %   1828   10.4 %
166 Alex 1.78                      : 1805   70  71    86    45.3 %   1837   11.6 %
167 Rival 1.14                     : 1800   82  83    70    46.4 %   1824    1.4 %
168 Eagle 0.7.5                    : 1791   79  80    71    44.4 %   1830    7.0 %
169 Amundsen 0.55                  : 1769   74  75    78    41.7 %   1827   11.5 %
170 Adamant 1.7                    : 1764   76  77    77    40.9 %   1828    9.1 %
171 J.S.BAM 0.52                   : 1755   74  75    78    39.7 %   1827   12.8 %
172 Kanguruh 1.92                  : 1745   80  83    71    36.6 %   1841    8.5 %

Yes,definitely around this Elo playing strength,a little bit less based on my tests....

[Graham Banks]

I'm rated 2032 fide and played two serious games with Smash and I lost both games 40moves 1hour.
Also I'm trying to rate it against several chessmaster 10 personalities and after 6 games it has a rating of 2180, I don't know of any MP version I'm using the single version, but it is by no means 1800elo.

Smash 1.0.3 is currently playing in my "Home On The Range" tournament which can be followed here:
http://kirr.homeunix.org/chess/discussi ... f=7&t=2913
The time control is CCRL 40/40.

The participants are:

Cheese 0.1c
Parrot 07.07.22
Alf 1.09
Mustang 4.97
Lime 62
Adam 3.1
BigLion 2.23x
OBender 3.1.0
Monarch 1.7
Marvin 1.3.0
Smash 1.0.3
Clueless 1.4

Regards, Graham.

[Dr.Wael Deeb]

Well, from 3 points out of 68 games it is totally impossible to determine a meaningful ELo rating. You might as well play zero games and throw dice. Whatever you get, is all statistical error.

No wonder there is no relationship whatsoever between the strength of the engine and the rating you give. Most of them have only 3 points out of a very large number of games. That means that most of the games must have been played against opponents so much stronger, that the outcome gives zero information.

If you want proof: accordng to yur ratings, Smash and TSCP are practically equally strong, and a 10-game match between them will end close to 5-5. But I can predict you that it will be more like 9-1 for Smash, if TSCP is lucky.

Please do run the match with TSCP and see for yourself....
In my tournaments,I play 33 engines two games against each other,round robin,that makes 1056 games in one tournament....
Smash has passed such a tournament where there were engines of various strength,but not that high as you expect,Rybka and Hiarcs were'nt there for sure....

[GS]

i am sure that it's installed correctly and plays with maximum strength,it's weak and it have played head to head with these engines at long time controls,it plays at the level of Nero

Sorry, something is very wrong then.
Smash will kill Nero _at least_ with 95/100 at any time control...
(of course even, if both are single cpu versions)

BTW if Smash had only 3/68 and Nero had 7/52 as your numbers
seem to indicate you already have introduced a big systematical error
in collecting the opponents.

Smash is in my amateur section in my rating list,the lowest division....

How is it possible all those programs mentioned for your lowest
divison have a _very low_ number of points?
That doesn't look like results of the same division at all, more like
a wrong collection of opponents.

Guenther

[Aleks Peshkov]

I want to add that Smash is open source program. It is written using nicely commented modern C++ style (however Smash architecture is not Object Oriented).

[Dr.Wael Deeb]

i am sure that it's installed correctly and plays with maximum strength,it's weak and it have played head to head with these engines at long time controls,it plays at the level of Nero

Sorry, something is very wrong then.
Smash will kill Nero _at least_ with 95/100 at any time control...
(of course even, if both are single cpu versions)

BTW if Smash had only 3/68 and Nero had 7/52 as your numbers
seem to indicate you already have introduced a big systematical error
in collecting the opponents.

Smash is in my amateur section in my rating list,the lowest division....

How is it possible all those programs mentioned for your lowest
divison have a _very low_ number of points?
That doesn't look like results of the same division at all, more like
a wrong collection of opponents.

Guenther

Hi Guenther,
As I mentioned before,they play in big tournaments with opponents having a various Elo strength,the games are not enough indeed,that's why I am not going to publish my rating list until I have a +200 games played by the engine....
This is an alpha stage of my rating list,just filtering the weak engines in a general division and more games are yet to be played....

[GS]

i am sure that it's installed correctly and plays with maximum strength,it's weak and it have played head to head with these engines at long time controls,it plays at the level of Nero

Sorry, something is very wrong then.
Smash will kill Nero _at least_ with 95/100 at any time control...
(of course even, if both are single cpu versions)

BTW if Smash had only 3/68 and Nero had 7/52 as your numbers
seem to indicate you already have introduced a big systematical error
in collecting the opponents.

Smash is in my amateur section in my rating list,the lowest division....

How is it possible all those programs mentioned for your lowest
divison have a _very low_ number of points?
That doesn't look like results of the same division at all, more like
a wrong collection of opponents.

Guenther

Hi Guenther,
As I mentioned before,they play in big tournaments with opponents having a various Elo strength,the games are not enough indeed,that's why I am not going to publish my rating list until I have a +200 games played by the engine....
This is an alpha stage of my rating list,just filtering the weak engines in a general division and more games are yet to be played....

Wael, you could do a much better selection already after a few games.
It is completely counter productive to let programs play >60 games
against opponents out of a margin of hope.
This adds zero info for any rating calculation, as HG already said.

Guenther