Sedat Canbaz

[Sedat Canbaz]

Just I'd like to add this as well,
For example, non-shared test (under Cutechess 1.3.1)
Artemis TR is lost several games on time... and now,
Under Cutechess 1.2.0, it seems all stable so far...

[Skynet]

One thing in SF-PB missing: not capable to use CTG..
But no one work is perfect and we've to be satisfied..

Yeah, well, SF-PB is a "continuation" in memory of Brainfish introduced by Thomas Zipproth at 2016, with support only for BIN format books and it is perfect in what it was designed for. There is plenty "identical" engines with CTG book support, just pick one and you're done.

[Sedat Canbaz]

Breaking News (06.10.2024):
1st of all, and for anyone missed: why Cutechess 1.2.0 GUI is used?
Due to it's best one / most stable that ever seen...proved /guaranteed!
E.g no any game is recorded to be lost on time...in other words: great!
And let's say what about other GUIs? not so bad too, but they are not
So suitable or much stable on this BIG bus (2x EPYC 7B12 128C / 256T)
What about Fritz GUI? because sometimes I get questions e.g why not
Used in SCCT, well, in the past.. I used Fritz GUIs and quite often.. but
We are in 2024 year, where unfortunately Fritz GUI is so poor too...sure
I refer in case of my side..known that no way of running parallel games!!
Sure it's possible..but too complicated running many Fritz GUIs at once
Where Cutechess GUI does this job perfectly.. and here the golden key..
Yes..in past, such as before 2000 years is alright..running 1 game...but..
Ok..and let's back to actual EvalFile CS ) not very often but sometimes
I organize these kind of competitions for example for previous: Results
And via current contest is already clear that their strength is quite close
But nn-86aaa607b31b is something that we should pay more attention!
Plus via this tour I was wondering also which Evalfiles are less drawish..
And after checking,. as we see: the lowest draws: 93%/ highest is 95%
Where B12tz.evL is still among the strongest plus least drawish...nice!
BIG thanks to all Evalfile authors, Skynet, Linrock..keep the good work!

The Winner: nn-86aaa607b31b.nnue - Congratulations to Linrock!!

For More Info/Final Standings etc:
https://sites.google.com/site/computers ... lfile-cs-5

GAMES:
https://mega.nz/file/C9IQiJ7a#KrP3H-FV_ ... hcXt7_5YNo

Greetings )

[Sedat Canbaz]

Hello Friends )

As usual, I am very pleased to announce:
Very soon new CS is coming, target: 3800+

And here are the planning Top players:

Artemis 2TR
Brainlearn 29
Corchess 170924
Fidelio 17 MPV
JigSaw 6.0
Killfish 280624
Kookaburra 3
MagicKnight
Marauders 3.9
Rems MPV Sep24
SF-Ivec 170924
SF-PB 280924
SF-PB 280924 SC
SF-POLY 210924a
SF-PRO 021024
ShashChess 36
SOLIDOR 180924
Stockfish 17
Stockfish 17 w32
Stockfish 041024

Btw, you may noticed or not:
I've included both former Champions: Killfish 280624, Kookaburra 3
And I wonder much what will be performance (by both Titans) now..
Stockfish 17 w32 will be in too.. let's see what will be Elo difference
And via this CS: several engines (by same author/s) will be played..
SF-PB 280924 SC = via nn-86aaa607b31b.nnue (Evalfile CS V's Winner)

If other new 3800+ Elo engine is missing just let me know please!
Sure if possible... to be created by known SF-based authors...or if
Nothing else to be available from known/trusted sites..but reminder:
My goal is simply that, I hope test mainly which never played in SCCT

Wishing good luck to all )

[Sedat Canbaz]

UPDATE

The tour is started and here is 1st Gauntlet entry by Stockfish 17 w32:
Reason of running 1st test in Gauntlet mode..for several reasons, e.g
Via this system way: 1st for stability, cause no idea what will be influence,
Since many years I did not touch 32-bit engine releases...2nd. in case of
Instability issues.. such as time forfeits, crashes etc. sure here I refer for
All engines.. just in case..these engines will be out.. otherwise, simply
Will be waste of time..and re-starting the tour from beginning is really
Pain...sure here I'm trying to share all my experiences with yıu..to gain...
You know we've to learn from each other...just for computer chess progress!

On other hand,
It's already clear that Stockfish 17 w32 will be one of last ranks..
But anyhow.. no problem at all... see you later )

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   0 Stockfish 17 w32              -66      22     154   40.6%   81.2% 
   1 SF-PRO 021024                 191     189       8   75.0%   50.0% 
   2 Rems MPV Sep24                137     147       8   68.8%   62.5% 
   3 Artemis 2TR                   137     147       8   68.8%   62.5% 
   4 Killfish 280624               102     134       7   64.3%   71.4% 
   5 ShashChess 36                  89     116       8   62.5%   75.0% 
   6 SF-PB 280924                   89     116       8   62.5%   75.0% 
   7 SF-Ivec 170924                 89     116       8   62.5%   75.0% 
   8 Marauders 3.9                  89     116       8   62.5%   75.0% 
   9 MagicKnight                    79     102       9   61.1%   77.8% 
  10 Fidelio 17 MPV                 70      92      10   60.0%   80.0% 
  11 Stockfish 041024               50      94       7   57.1%   85.7% 
  12 SF-PB 280924 SC                50      94       7   57.1%   85.7% 
  13 Kookaburra 3                   50      94       7   57.1%   85.7% 
  14 Stockfish 17                   44      82       8   56.3%   87.5% 
  15 SOLIDOR 180924                 44      82       8   56.3%   87.5% 
  16 SF-POLY 210924a                 0       0       9   50.0%  100.0% 
  17 JigSaw 6.0                      0       0       8   50.0%  100.0% 
  18 Corchess 170924                 0       0       9   50.0%  100.0% 
  19 Brainlearn 29                   0       0       9   50.0%  100.0% 

154 of 1026 games finished.

Conditions:
2x Epyc 7B12, CuteChess120, 1 Core, Ponder OFF, 30s+05s, Balsa/Unique, 64 Hash, 5-MEN

[Sedat Canbaz]

UPDATE 2

Well, here is 2nd update for today.. but this time the players in round-robin:
Note also that Stockfish 17 w32 test ended, but will be merged/shared later..

The Current Leader: Corchess 170924 - My Congratulations to Ivec !
Btw, Kookaburra 3 is even as old dated but Champion is Champion..
As we see still among best ones! like GOLD 24C, never loses worth!)

About SF-PB 280924 SC,
I never play for 2nd place too ) but we will see.. still too many games
Waiting us..who knows?) maybe in next rounds SC will be next Leader..

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 Corchess 170924                11       9     407   51.6%   92.9% 
   2 Brainlearn 29                   6       7     403   50.9%   96.3% 
   3 SF-PB 280924 SC                 4       7     402   50.6%   95.8% 
   4 Kookaburra 3                    3       7     404   50.5%   95.5% 
   5 SF-PB 280924                    3       8     403   50.4%   93.8% 
   6 Rems MPV Sep24                  3       8     406   50.4%   93.8% 
   7 Stockfish 17                    2       9     404   50.2%   93.1% 
   8 SF-Ivec 170924                  2       8     405   50.2%   94.6% 
   9 Fidelio 17 MPV                  1       9     402   50.1%   93.3% 
  10 MagicKnight                     1       9     404   50.1%   92.3% 
  11 Killfish 280624                 1       9     405   50.1%   92.3% 
  12 SOLIDOR 180924                  0       7     405   50.0%   95.6% 
  13 SF-PRO 021024                  -2       8     405   49.8%   94.1% 
  14 Artemis 2TR                    -2       8     403   49.8%   94.0% 
  15 Marauders 3.9                  -2       8     402   49.8%   95.0% 
  16 Stockfish 041024               -3       7     404   49.6%   95.8% 
  17 SF-POLY 210924a                -7       9     404   49.0%   93.1% 
  18 JigSaw 6.0                     -7       8     404   49.0%   94.6% 
  19 ShashChess 36                 -15       9     404   47.9%   92.3% 

3838 of 9234 games finished.

Conditions:
2x Epyc 7B12, CuteChess120, 1 Core, Ponder OFF, 30s+05s, Balsa/Unique, 64 Hash, 5-MEN

[Sedat Canbaz]

UPDATE 3

One more entry: C-Dragon v3.0 by Tanick Ramz (OrgZ)
For anyone missed, it's one of Former Champions too!

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   0 C-Dragon v3.0                   5      10     305   50.7%   93.4% 
   1 Stockfish 041024               50      66      14   57.1%   85.7% 
   2 Fidelio 17 MPV                 22      42      16   53.1%   93.8% 
   3 Brainlearn 29                  22      42      16   53.1%   93.8% 
   4 Artemis 2TR                    22      42      16   53.1%   93.8% 
   5 Stockfish 17                    0       0      14   50.0%  100.0% 
   6 SOLIDOR 180924                  0       0      14   50.0%  100.0% 
   7 ShashChess 36                   0      70      14   50.0%   85.7% 
   8 SF-PRO 021024                   0       0      16   50.0%  100.0% 
   9 SF-POLY 210924a                 0       0      16   50.0%  100.0% 
  10 SF-PB 280924 SC                 0       0      16   50.0%  100.0% 
  11 SF-Ivec 170924                  0       0      15   50.0%  100.0% 
  12 Rems MPV Sep24                  0       0      16   50.0%  100.0% 
  13 Marauders 3.9                   0       0      16   50.0%  100.0% 
  14 MagicKnight                     0       0      16   50.0%  100.0% 
  15 Kookaburra 3                    0       0      15   50.0%  100.0% 
  16 Corchess 170924               -22      42      16   46.9%   93.8% 
  17 SF-PB 280924                  -23      44      15   46.7%   93.3% 
  18 Killfish 280624               -23      79      15   46.7%   80.0% 
  19 JigSaw 6.0                    -44      85      16   43.8%   75.0% 
  20 Stockfish 17 w32             -110     100      13   34.6%   69.2% 

305 of 1080 games finished.

Note also that,
Very soon I hope to share all the played tournament results...

By the way, I have good news so far, no any crash, time loss etc..
It seems Cutechess v1.2.0 plays as serious BIG role over ....
Just imagine... if the current tour was under Cutechess 1.3.1...
But it is not... )

Greetings

[Sedat Canbaz]

UPDATE 4

Well, more Champions 2024 are planning to be played,
Sure the Top ones, which are close to 3800+ Elo points..

In other words,
In the end, each player will contain min 1500+ games
In this way, I expect to see less 'error margin' as well
Otherwise..you know already..running small number,
Such as 100-300-500 games is lottery to see the best!

Code: Select all

Corrchess 071223      I. Ivec                XLVII
Kookaburra 2          Dorsz                  XLVII
Killfish 270224       Anton Ces              XLVII
Killfish 190324       Anton Ces              XLIX
Stockfish 16.1        Stockfish Team         L
SF-Poly 020624        Tanick Ramz            LI 2nd
SF-PB 120624          Skynet (Alex)          LII
Stockfish 220923      SF team                Medium Size I

Source:
https://sites.google.com/site/computers ... eng-champs

[Sedat Canbaz]

UPDATE 5

1st of all,
Who said that latest SOLIDOR is weaker than real mother..?)
Why I say like that? recently the SOLIDOR's author claim that
Latest Stockfish is best... maybe this is true or maybe not..
After all, time will tell...

And as before,
Corchess 170924 leads.. impressive..just behind is Kookaburra 3,
In 3rd rank is SOLIDOR 180924 not so bad too...

Code: Select all

Rank Name               Elo    +    - games score oppo. draws 
   1 Corchess 170924   3817    4    4  1080   52%  3808   92% 
   2 Kookaburra 3      3815    4    4  1080   51%  3808   94% 
   3 SOLIDOR 180924    3814    4    4  1080   51%  3808   94% 
   4 Rems MPV Sep24    3813    4    4  1080   51%  3808   92% 
   5 MagicKnight       3813    4    4  1080   51%  3808   92% 
   6 Fidelio 17 MPV    3813    4    4  1080   51%  3808   94% 
   7 SF-PB 280924 SC   3813    4    4  1080   51%  3808   93% 
   8 SF-PB 280924      3813    4    4  1080   51%  3808   93% 
   9 C-Dragon v3.0     3812    4    4  1080   51%  3808   93% 
  10 Stockfish 17      3812    4    4  1080   51%  3808   93% 
  11 Brainlearn 29     3812    4    4  1080   51%  3808   94% 
  12 SF-Ivec 170924    3811    4    4  1080   51%  3808   93% 
  13 Killfish 280624   3810    4    4  1080   50%  3808   92% 
  14 Stockfish 041024  3809    4    4  1080   50%  3808   93% 
  15 Artemis 2TR       3809    4    4  1080   50%  3808   93% 
  16 SF-PRO 021024     3809    4    4  1080   50%  3808   92% 
  17 Marauders 3.9     3808    4    4  1080   50%  3809   93% 
  18 SF-POLY 210924a   3807    4    4  1080   50%  3809   94% 
  19 JigSaw 6.0        3803    4    4  1080   49%  3809   93% 
  20 ShashChess 36     3802    4    4  1080   49%  3809   92% 
  21 Stockfish 17 w32  3763    3    3  1080   37%  3811   74% 

[Sedat Canbaz]

UPDATE 6

1st. according to current SCCT (30s+0.5s) tour,
SF17 is about +2 Elo better than SF 041024 dev!
What does it mean? sure that means in my eyes:
Not always as newer better...! sad but true..!

Code: Select all

Rank Name               Elo    +    - games score oppo. draws 
  11 Stockfish 17      3811    4    4  1512   51%  3807   94% 
  17 Stockfish 041024  3809    4    4  1512   50%  3807   94% 

2nd NON-SCCT, according to abrok.eu:
https://abrok.eu/stockfish/?page=1

E.g based on theirs info.in almost every Stockfish dev releases..
They are mentioning serious Elo+ numbers, such as improvements!
Is this can be true ? maybe on their tests...at least there is one true:
Theirs mentioned Elo+ points do not reflect to reality (in SCCT)!
What I can suggest to SF team:
Start using better /stronger opening lines...with better quality games..
And then we can compare...otherwise, what they do: is waste of time!

Btw,
Soon I hope to share the latest full rankings
On SedatChess site..so please stay tuned )

Greetings