SCCT - Blitz Rating List 2015

[Sedat Canbaz]

Hello There,


Last Updates:
Android engines (as 1 core + ponder on) played on my Samsung Galaxy S4:
DiscoCheck 5.2.1 Droid
Gaviota 1.0 Droid
Komodo 8 Droid
Texel 1.05a8 Droid
Senpai 1.0 Droid
Stockfish 121014 Droid

Soon more games will be added...


Best,
Sedat

[Sedat Canbaz]

Btw,
I've just checked more carefully the time losses played in my current rating list

And here are the chess engines, which frequently lose on time:

Note: Lower Better !

Komodo 1318 x64: 119 loses on time
Djinn 1.008 x64: 92 loses on time
Phalanx XXIII JA: 75 loses on time
Fire 4 x64: 58 loses on time
Spark 1.0 x64: 57 loses on time
Spike 1.4: 36 loses on time
Stockfish 110115 x64: 35 loses on time
Chiron x64: 35 loses on time
RedQueen 1.0.0 x64: 31 loses on time
Senpai 1.0 x64: 29 loses on time
Zappa Mexico II x64: 28 loses on time
Dirty 25082011 x64: 28 loses on time
Octochess r5132 JA x64: 27 loses on time
NirvanaChess 2.0 x64: 26 loses on time
Francesca MAD 0.19: 26 loses on time
Stockfish 221214 x64: 25 loses on time
DiscoCheck 5.2 x64: 24 loses on time
Glass 2 x64: 24 loses on time
Glaurung 2.2 x64: 22 loses on time
Gull 3 x64: 15 loses on time
Booot 5.1.0: 15 loses on time
Hannibal 1.4b x64: 11 loses on time
Yace 0.99.87: 11 loses on time
iCE 1.0 v1619 x64: 10 loses on time


Note also that I did not include those time losses in the current Elo calculation
Otherwise, the standings would be quite different ...
Then Houdini 4 would be rated as 1st or 2nd )!

And I hope we will see less time forfeits in the next new engine releases )!


Best,
Sedat

[Sedat Canbaz]

And here is the direct link of current Android engine rankings:
http://www.sedatcanbaz.com/chess/?page_id=1339

[Sedat Canbaz]

Last Updates:
Arasan 17.5 x64 (approx. +60 Elo over v 17.2)
Texel 1.05 x64 (approx. +40 Elo over v 1.04)

Great...and well-done to both engine programmers !)

For Current Standings:
http://www.sedatcanbaz.com/chess/?page_id=1339

And since a few days,
I switched again to fisher time control (3min+1sec)

Note also that,
This time I decided to include all games, which were lost on time...
Because I don't want to lose the traditionality of my rules )!
And now the standings are slightly changed...
Just one example: Stockfish 210614 x64 is rated better than Stockfish 221214 x64
Normally without calculating the time losses, Stockfish 221214 x64 was approx. 10 Elo stronger...
But what a pitty this issue happened due to Stockfish 221214 x64 has many games lost on time

But however,
I think this is the right way, otherwise the programmers will never see their mistakes )!

Btw,
Komodo 1318 x64 is removed from current participation, because it's a private engine...
In other words, I plan to test only the public versions in the current competition
And I already started testing Komodo 8 x64 ...


Greetings,
Sedat

[Frank Quisinsky]

In other words, I plan to test only the public versions in the current competition
And I already started testing Komodo 8 x64 ...

...


That is the best way for a rating list other can look. Private available engines are not very interesting, other cann't have this material.

Good!

Best
Frank

[lkaufman]

Btw,
I've just checked more carefully the time losses played in my current rating list

And here are the chess engines, which frequently lose on time:

Note: Lower Better !

Komodo 1318 x64: 119 loses on time
Djinn 1.008 x64: 92 loses on time
Phalanx XXIII JA: 75 loses on time
Fire 4 x64: 58 loses on time
Spark 1.0 x64: 57 loses on time
Spike 1.4: 36 loses on time
Stockfish 110115 x64: 35 loses on time
Chiron x64: 35 loses on time
RedQueen 1.0.0 x64: 31 loses on time
Senpai 1.0 x64: 29 loses on time
Zappa Mexico II x64: 28 loses on time
Dirty 25082011 x64: 28 loses on time
Octochess r5132 JA x64: 27 loses on time
NirvanaChess 2.0 x64: 26 loses on time
Francesca MAD 0.19: 26 loses on time
Stockfish 221214 x64: 25 loses on time
DiscoCheck 5.2 x64: 24 loses on time
Glass 2 x64: 24 loses on time
Glaurung 2.2 x64: 22 loses on time
Gull 3 x64: 15 loses on time
Booot 5.1.0: 15 loses on time
Hannibal 1.4b x64: 11 loses on time
Yace 0.99.87: 11 loses on time
iCE 1.0 v1619 x64: 10 loses on time


Note also that I did not include those time losses in the current Elo calculation
Otherwise, the standings would be quite different ...
Then Houdini 4 would be rated as 1st or 2nd )!

And I hope we will see less time forfeits in the next new engine releases )!


Best,
Sedat

Thanks for pointing out that Komodo is losing a lot of games on time at game/3'. I expect we won't lose any on time with an increment. What this tells me is that our default value for "Move Overhead Milliseconds" of 10 is too low for Arena (and I also know it's too low for LittleBlitzer), so I suggest people who play without increment should raise it to 20 or 30, whatever will avoid the time losses. I think we'll raise it for Komodo 9 to avoid this issue. In any case you are doing the right thing to test with increment, that avoids the whole issue of time forfeits or of having to play instantaneous moves in super-long games.

[Ferdy]

Last Updates:
Arasan 17.5 x64 (approx. +60 Elo over v 17.2)
Texel 1.05 x64 (approx. +40 Elo over v 1.04)

Great...and well-done to both engine programmers !)

For Current Standings:
http://www.sedatcanbaz.com/chess/?page_id=1339

And since a few days,
I switched again to fisher time control (3min+1sec)

Note also that,
This time I decided to include all games, which were lost on time...
Because I don't want to lose the traditionality of my rules )!
And now the standings are slightly changed...
Just one example: Stockfish 210614 x64 is rated better than Stockfish 221214 x64
Normally without calculating the time losses, Stockfish 221214 x64 was approx. 10 Elo stronger...
But what a pitty this issue happened due to Stockfish 221214 x64 has many games lost on time

But however,
I think this is the right way, otherwise the programmers will never see their mistakes )!

Btw,
Komodo 1318 x64 is removed from current participation, because it's a private engine...
In other words, I plan to test only the public versions in the current competition
And I already started testing Komodo 8 x64 ...


Greetings,
Sedat

Thanks for doing these tests. I have seen in the list,

Code: Select all

52 Deuterium 14 x64           : 2904.2     445.5    1131   39.4%

Note I have released 3 versions in year 2014 (code 14), namely,
* Deuterium v14.1.32.119
* Deuterium v14.2.33.276
* Deuterium v14.3.34.130

The latest version is Deuterium v14.3.34.130. You may use (a shorter name) Deuterium 14.3 x64 to properly identify uniquely each playing version. I am more interested on the games played by v14.3 at the moment, to possibly detect some weaknesses, and make perhaps some changes to address those weaknesses. This is ponder on so the games here are interesting as I am not testing it seriously on ponder on games.

The games at the moment are still not available.

[Sedat Canbaz]

In other words, I plan to test only the public versions in the current competition
And I already started testing Komodo 8 x64 ...

...


That is the best way for a rating list other can look. Private available engines are not very interesting, other cann't have this material.

Good!

Best
Frank

Hello Frank,


Yes...that's right and it's better to test the public chess engines...
But however, once more I'd like say 'Thanks' to Komodo team
They give me a chance to include Komodo 1318 in my recent tournaments
And I was quite exiting testing it...

Btw,
I can't say the same thing about testing opening books
For example, SCCT Book Tournaments are based mainly on private books

And in this way, the book competitions are getting more exiting

Otherwise,
A experienced book maker can look at the available books and can tune/optimize to play stronger...
But when the book participants are private, then the situation is different
Then only the the new very strong opening move ideas/lines that never played are usually at one of the top places )!

In shortly, (If the book tournaments include the private books too)
Then the public books has no BIG chances to be rated at one of the top places!

Best,
Sedat

[Sedat Canbaz]

Thanks for pointing out that Komodo is losing a lot of games on time at game/3'. I expect we won't lose any on time with an increment. What this tells me is that our default value for "Move Overhead Milliseconds" of 10 is too low for Arena (and I also know it's too low for LittleBlitzer), so I suggest people who play without increment should raise it to 20 or 30, whatever will avoid the time losses. I think we'll raise it for Komodo 9 to avoid this issue. In any case you are doing the right thing to test with increment, that avoids the whole issue of time forfeits or of having to play instantaneous moves in super-long games.

Hello Larry,

Not at all...

Interesting, and I did not know about "Move Overhead Milliseconds"
But its ok...as you said with 3min + 1sec probably we will see less time forfeits by mostly of the engines

Btw,
After 952 games, Komodo 8 is doing very good so far...
Just 18 Elo behind than recent Stockfish 110115
And my strong 6th sense says that Komodo 9 can top SCCT )! even at blitz ...


Best,
Sedat

[Sedat Canbaz]

Thanks for doing these tests. I have seen in the list,

Code: Select all

52 Deuterium 14 x64           : 2904.2     445.5    1131   39.4%

Note I have released 3 versions in year 2014 (code 14), namely,
* Deuterium v14.1.32.119
* Deuterium v14.2.33.276
* Deuterium v14.3.34.130

The latest version is Deuterium v14.3.34.130. You may use (a shorter name) Deuterium 14.3 x64 to properly identify uniquely each playing version. I am more interested on the games played by v14.3 at the moment, to possibly detect some weaknesses, and make perhaps some changes to address those weaknesses. This is ponder on so the games here are interesting as I am not testing it seriously on ponder on games.

The games at the moment are still not available.

Hello Ferdinand,

You are welcome, it's my pleasure

Yes...it's your latest official version
That's right.... I edited the name manually
To be honest, due to I don't like very much many many long numbers ))

BTW,
SCCT's games are already available !)

And it all depends on my free time,
Usually I upload all played games to the server once a month

Best,
Sedat