SedatChess

[Sedat Canbaz]

Hello,

I have rewritten your news for my better understanding as follows:

Hello )

Very interesting..thank you very much for your efforts ) and to be honest,
I never would have guessed that some of my original news will be derived!

On the other hand,
I was quite clear in LII's Breaking News, but I admit that
Many of your derived comments, words etc. are enhanced...
And it seems your English is good, better than mine for sure!
But this is also true that my English is not so bad too ) since
You managed to understand..sure not all, but mostly my words..
Note also that my English is not native so I make mistakes...
Not so frequent..but I make..for example, I've learned my
English language via practice (conversations, dictionaries..)
Btw, normally I speak 4 (four) languages more, I mean that
When you speak multiple languages..simply no way to speak
All of them perfectly...as same words are going for books,
Engines, authors, testers etc e.g who/which one is perfect?)

Meanwhile, not all of your derived comments are correct!
Very likely, you did not understand what I was meaning...
Besides, you missed to mentions some of my statements too,
Sure I'm not going to count all, but at least I can give just
Only one example: where I was referring simply about that
The bicycle's wheel goes forward, not backward...
Where you wrote about moves..yes, normally the chess pieces
Moves can go backward in game (exception pawns), right ?)
But I never noticed the bicycle's wheel to go back...
Sure I am meaning in driving...

Greetings )

[Sedat Canbaz]

Hello there,

As usual, I am pleased to announce that
Here is latest Simex 2's Similarity Report,
Played on my AMD Ryzen 7 4800H machine

What is new,
Instead of Opening.epd, Simex.epd is used,
Plus all are tested at 3sec/per pos. 1 Core etc.
E.g via ultra fast TC, such as 100ms is not much
Effective (as before...) with some of latest SF
Based engines...but if giving 3 seconds, 1 Core
Then as you may see too, the ones, which are
Based on same source, theirs similarity results
Are very close to each other..and as clear proof,
You can check SF-PB 120624 vs SF 120634 is 98.99%
Other good example:
Cfish 030820 vs SF310720: 97.98%

Btw, a SURPRISE ) and what is going on ?)
Other new SF-based one with 97.98% in similarity e.g
Latest Killfish 280624 is highly close to SF 270624 dev.

As Final Conclusion,
Over long past years... I've tested a lot of chess engines, using
Various Calculation, Similarity tools etc. sure just for accurate
Rankings, Sim results etc. In other words, I wish to say that
It is time to stop using the Similarity tools, as main reason:
The originality is almost dead...sure if we are looking, testing
The latest Top engines..at least quite frequently, the latest
SIM results confirm this... otherwise, all my free efforts
Will be simply waste of my time...Besides, I have no much
Free time to check each engine..before starting new tours!
Even so.. I mean just in case.. each time I have to increase
The similarity rule limit at higher and higher...SORRY..
And thanks for your understanding!

To be more clear, as far as possible,
I do not plan to use anymore the similarity tools...
I am aware too that without 'similarity rule'
The next new tour cond. will be not much fair,
But on the other hand, who cares?

And here is the latest Similarity Results:



Download Full Standings:
https://mega.nz/file/y5IjBSJD#Otg4kVhlK ... pG_zk1zGOM

Best,
Sedat

[OrgZ]

Hello there,

As usual, I am pleased to announce that
Here is latest Simex 2's Similarity Report,
Played on my AMD Ryzen 7 4800H machine

What is new,
Instead of Opening.epd, Simex.epd is used,
Plus all are tested at 3sec/per pos. 1 Core etc.
E.g via ultra fast TC, such as 100ms is not much
Effective (as before...) with some of latest SF
Based engines...but if giving 3 seconds, 1 Core
Then as you may see too, the ones, which are
Based on same source, theirs similarity results
Are very close to each other..and as clear proof,
You can check SF-PB 120624 vs SF 120634 is 98.99%
Other good example:
Cfish 030820 vs SF310720: 97.98%

Btw, a SURPRISE ) and what is going on ?)
Other new SF-based one with 97.98% in similarity e.g
Latest Killfish 280624 is highly close to SF 270624 dev.

As Final Conclusion,
Over long past years... I've tested a lot of chess engines, using
Various Calculation, Similarity tools etc. sure just for accurate
Rankings, Sim results etc. In other words, I wish to say that
It is time to stop using the Similarity tools, as main reason:
The originality is almost dead...sure if we are looking, testing
The latest Top engines..at least quite frequently, the latest
SIM results confirm this... otherwise, all my free efforts
Will be simply waste of my time...Besides, I have no much
Free time to check each engine..before starting new tours!
Even so.. I mean just in case.. each time I have to increase
The similarity rule limit at higher and higher...SORRY..
And thanks for your understanding!

To be more clear, as far as possible,
I do not plan to use anymore the similarity tools...
I am aware too that without 'similarity rule'
The next new tour cond. will be not much fair,
But on the other hand, who cares?

And here is the latest Similarity Results:



Download Full Standings:
https://mega.nz/file/y5IjBSJD#Otg4kVhlK ... pG_zk1zGOM

Best,
Sedat

Yes the simex tool has its flaw. But it seems you didn't understand me here
viewtopic.php?f=7&t=4074&p=36250#p36250

If you did you wouldn't be surprised

[OrgZ]

Hello there,

As usual, I am pleased to announce that
Here is latest Simex 2's Similarity Report,
Played on my AMD Ryzen 7 4800H machine

What is new,
Instead of Opening.epd, Simex.epd is used,
Plus all are tested at 3sec/per pos. 1 Core etc.
E.g via ultra fast TC, such as 100ms is not much
Effective (as before...) with some of latest SF
Based engines...but if giving 3 seconds, 1 Core
Then as you may see too, the ones, which are
Based on same source, theirs similarity results
Are very close to each other..and as clear proof,
You can check SF-PB 120624 vs SF 120634 is 98.99%
Other good example:
Cfish 030820 vs SF310720: 97.98%

Btw, a SURPRISE ) and what is going on ?)
Other new SF-based one with 97.98% in similarity e.g
Latest Killfish 280624 is highly close to SF 270624 dev.

As Final Conclusion,
Over long past years... I've tested a lot of chess engines, using
Various Calculation, Similarity tools etc. sure just for accurate
Rankings, Sim results etc. In other words, I wish to say that
It is time to stop using the Similarity tools, as main reason:
The originality is almost dead...sure if we are looking, testing
The latest Top engines..at least quite frequently, the latest
SIM results confirm this... otherwise, all my free efforts
Will be simply waste of my time...Besides, I have no much
Free time to check each engine..before starting new tours!
Even so.. I mean just in case.. each time I have to increase
The similarity rule limit at higher and higher...SORRY..
And thanks for your understanding!

To be more clear, as far as possible,
I do not plan to use anymore the similarity tools...
I am aware too that without 'similarity rule'
The next new tour cond. will be not much fair,
But on the other hand, who cares?

And here is the latest Similarity Results:



Download Full Standings:
https://mega.nz/file/y5IjBSJD#Otg4kVhlK ... pG_zk1zGOM

Best,
Sedat

Yes the simex tool has its flaw, due to nnue. But it seems you didn't understand me here
viewtopic.php?f=7&t=4074&p=36250#p36250

If you did you wouldn't be surprised, look at kookaburra and jigsaw against Stockfish

[Sedat Canbaz]

Yes the simex tool has its flaw. But it seems you didn't understand me here
viewtopic.php?f=7&t=4074&p=36250#p36250

If you did you wouldn't be surprised

Just a moment,
I am not perfect...and I never promised to
Anyone here that I understand everything

Anyhow, can you be more specific
About where I do not understand ?)
At least, if you tell.. I can try later...
Thanks in advance )

About Simex 2 program,
It may be has flaws...but on other hand, no doubt that
It's a good and useful tool, where rarely makes mistakes!

Greetings )

[OrgZ]

Dear friend, perhaps you are viewing this the wrong way, according to the way you are responding. you are being surprised by the high similarity between the two engines. which i have covered already on what's going on. . perhaps if you could utilize a tool called Winmerge you could be able to see the similarities and differences in the source codes, even though you wont be able to tell what this does(whatever is within the code) but atleast you will see why such a big similarity


Speaking of similarity, am working on a tool to do that. so far it can detect % scores very well but still needs work. i tried it on Raid.
firstly i made two copies of Raid and renamed the other one to Raid _avx2, while the other one remained as Raid v3.5_avx2. similarity report was 100%
then i went to the source and changed the name, book naming and author inside and recompiled and i got 98.99%
then now for the ultimate test, i took stockfish 240615 and Compared it with raid and it gave 5.69%. this is true in a way (though the percentage is very low) because Raid v3.5 is very old december code!

And my tool is very simple, you just place any number of engines inside the folder and double click the Similarity.exe file and it checks within seconds its done, it will generate Results.txt

i also took stockfish 240612 and SF-PB 120624 98.99% report (there was no need for me to test this bcz i already know the difference between the two engines, but i did it to test if it can really work on all engines)

but it still needs refining, but so far i have rested from it. one day i will get back to it again and refine it more.

[Sedat Canbaz]

Dear friend, perhaps you are viewing this the wrong way, according to the way you are responding. you are being surprised by the high similarity between the two engines. which i have covered already on what's going on. . perhaps if you could utilize a tool called Winmerge you could be able to see the similarities and differences in the source codes, even though you wont be able to tell what this does(whatever is within the code) but atleast you will see why such a big similarity


Speaking of similarity, am working on a tool to do that. so far it can detect % scores very well but still needs work. i tried it on Raid.
firstly i made two copies of Raid and renamed the other one to Raid _avx2, while the other one remained as Raid v3.5_avx2. similarity report was 100%
then i went to the source and changed the name, book naming and author inside and recompiled and i got 98.99%
then now for the ultimate test, i took stockfish 240615 and Compared it with raid and it gave 5.69%. this is true in a way (though the percentage is very low) because Raid v3.5 is very old december code!

And my tool is very simple, you just place any number of engines inside the folder and double click the Similarity.exe file and it checks within seconds its done, it will generate Results.txt

i also took stockfish 240612 and SF-PB 120624 98.99% report (there was no need for me to test this bcz i already know the difference between the two engines, but i did it to test if it can really work on all engines)

but it still needs refining, but so far i have rested from it. one day i will get back to it again and refine it more.

I see...

But it seems you did not read well or you have difficulties
Regarding what I wrote in my recent main posting....
In short,
I've lost my interest using similarity tools anymore!
About 25 years is enough of checking eng originalities
And wishing you good luck with your new sim projects...

Greetings

[OrgZ]

[Sedat Canbaz]

Hello Chess Friends,

Just another new CS (1m+06s) is started..
Main Target: Testing 3700+ Elo Engines!
1 (One) Chess Engine per Author/s and
As players: Publicly available are allowed..

What is new (via this competition),
Exception that there are many 3750+ participants..
A newer Unique book vers used, which is optimized
To prefer Whites plus Blacks wins, in other words:
Any opening contains wins of 3750+ (by both sides)
E.g older Unique vers were tuned of wins by 3700+

And here are latest LIVE standings, where Kookaburra,
Killfish, SFNNv6, Brainlearn..are with impressive start!
Sure the Championship is just in beginning...let's see
After 1100+ games (per player)...BTW, I wonder much
This time SF-PB's secret weapon will work or not..?!
After all, time will tell ...

GOOD LUCK TO ALL )

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Kookaburra 3                   44      16     214   56.3%   86.4% 
   2 Killfish 280624                37      15     215   55.3%   88.4% 
   3 SFNNv6 16.1                    36      18     213   55.2%   85.0% 
   4 Brainlearn 28.1                35      16     212   55.0%   87.3% 
   5 D.Blue 180524                  33      16     214   54.7%   87.9% 
   6 Corchess 180624                31      15     215   54.4%   89.3% 
   7 Stockfish 230624               28      15     213   54.0%   89.2% 
   8 SF-PB 120624                   25      15     211   53.6%   89.1% 
   9 JigSaw v5.9                    21      15     215   53.0%   89.3% 
  10 Beast 15                       20      16     214   52.8%   87.9% 
  11 Yuliirma 2.1                   18      16     213   52.6%   88.3% 
  12 ProteusSF Aureo                16      16     213   52.3%   87.8% 
  13 XTD 010723                     15      18     213   52.1%   85.4% 
  14 Polyfish 140124                 6      14     214   50.9%   90.7% 
  15 Crystal 8                       2      15     212   50.2%   90.1% 
  16 Kayra 1.7                       2      18     213   50.2%   85.4% 
  17 CatroP 310723                   2      13     214   50.2%   92.1% 
  18 SF Polyglot 15.1               -2      16     212   49.8%   88.2% 
  19 Nebula-Vortex                  -2      15     212   49.8%   90.1% 
  20 Fisherov M 1.2                 -3      17     215   49.5%   87.0% 
  21 BigBallz 010922               -10      14     214   48.6%   90.7% 
  22 Zeus 26.5                     -13      16     214   48.1%   87.9% 
  23 Fat Titz 2                    -13      16     214   48.1%   88.8% 
  24 Raubfisch X48e                -26      18     214   46.3%   84.1% 
  25 K.Dragon 3.3                  -26      19     214   46.3%   82.2% 
  26 Berserk 13                    -36      20     213   44.8%   81.2% 
  27 Honey v14.1.01                -43      17     212   43.9%   84.9% 
  28 Obsidian 12.29                -49      21     214   43.0%   78.5% 
  29 Cfish 120821                  -59      22     213   41.5%   76.5% 
  30 Caissa v1.19                  -89      24     215   37.4%   71.2% 

3202 of 17400 games finished.

And please stay tıned,
More details, updates soon as possible..


Best,
Sedat

[Sedat Canbaz]

UPDATE 2

Wow..JigSaw v5.9 is Current Leader - Congrats to Tanick Ramz !
Where SF-PB 120624 is 2nd and Kookaburra3 is 3rd for right now,
And Killfish 280624 is currently 4th, another serious opponent..

Note also that,
After the end of this CS...I plan to organize another tourney
With the highest ranked 4 (four) engines, which will be played
Under better optimal tour conditions...

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 JigSaw v5.9                    37       9     672   55.3%   86.8% 
   2 SF-PB 120624                   34       9     670   54.9%   87.9% 
   3 Kookaburra 3                   33       8     670   54.7%   89.1% 
   4 Killfish 280624                32       9     671   54.5%   87.9% 
   5 SFNNv6 16.1                    30       9     669   54.3%   86.7% 
   6 Corchess 180624                29       9     669   54.1%   87.9% 
   7 Stockfish 230624               28       9     668   54.0%   88.9% 
   8 D.Blue 180524                  28       9     670   54.0%   87.8% 
   9 Brainlearn 28.1                28       9     670   54.0%   89.0% 
  10 Beast 15                       23       9     671   53.3%   87.8% 
  11 Yuliirma 2.1                   21       8     671   53.0%   89.6% 
  12 XTD 010723                     20       9     671   52.9%   87.6% 
  13 Polyfish 140124                18       9     670   52.6%   88.5% 
  14 ProteusSF Aureo                14       9     669   51.9%   88.9% 
  15 Crystal 8                       5       9     669   50.7%   87.7% 
  16 SF Polyglot 15.1                2       9     668   50.3%   88.6% 
  17 CatroP 310723                  -3       9     667   49.6%   88.5% 
  18 Nebula-Vortex                  -3       9     670   49.6%   87.8% 
  19 Kayra 1.7                      -7       9     670   49.0%   88.5% 
  20 Zeus 26.5                      -8      10     668   48.9%   86.7% 
  21 Fisherov M 1.2                 -8       9     671   48.8%   87.5% 
  22 BigBallz 010922               -17       9     668   47.6%   88.0% 
  23 Raubfisch X48e                -23      10     669   46.7%   86.2% 
  24 Fat Titz 2                    -23       9     671   46.6%   88.5% 
  25 K.Dragon 3.3                  -35      11     671   44.9%   80.9% 
  26 Honey v14.1.01                -36      11     671   44.9%   82.0% 
  27 Berserk 13                    -38      11     668   44.5%   81.0% 
  28 Obsidian 12.29                -49      11     670   43.1%   80.4% 
  29 Cfish 120821                  -54      11     669   42.3%   79.5% 
  30 Caissa v1.19                  -79      13     671   38.8%   74.4% 

10046 of 17400 games finished.