Chess Forum

Posted: **Sat Jun 22, 2024 10:40 am**

Update

Impressive! Private 19Sf is 1st qualified player (by Tanick Ramz engines)!

1   Private 19Sf    51.0 - 49.049.5 - 50.549.5 - 50.552.0 - 48.051.0 - 49.0**      253.0/500
2   SF-POLY 020624  49.0 - 51.051.0 - 49.049.5 - 50.552.0 - 48.049.5 - 50.5 **     251.0/500
3   C-Dragon v3.1   50.5 - 49.549.0 - 51.050.5 - 49.548.0 - 52.052.5 - 47.5  **    250.5/500
4   JigSaw v5.8     50.5 - 49.550.5 - 49.549.5 - 50.549.5 - 50.549.0 - 51.0   **   249.0/500 
5   Marauders 3.6   48.0 - 52.048.0 - 52.052.0 - 48.050.5 - 49.550.5 - 49.5    **  249.0/500 
6   C-Dragon v3.0   49.0 - 51.050.5 - 49.547.5 - 52.551.0 - 49.049.5 - 50.5     ** 247.5/500

As next Candidate tour:
Eduard Nemeth's engines will be played each other..
Good Luck to Mr. Eduard )

Greetings

Posted: **Sat Jun 22, 2024 10:56 am**

UPDATE 2

Another great score! LittleBeast 15 is next qualified..(by Eduard Nemeth)!

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 LittleBeast 15                  6      10     500   50.8%   88.4% 
   2 Deep Ripper                     6      11     500   50.8%   86.4% 
   3 Big-SF 110324                  -2      11     500   49.7%   85.8% 
   4 Sun_15                         -3      12     500   49.6%   85.2% 
   5 SFS 130424                     -3      11     500   49.6%   87.2% 
   6 Steel Fighter                  -3      11     500   49.5%   87.8% 

1500 of 1500 games finished.

Note also that (for determining the Winner...),
There was Tie-Break Match, LittleBeast 15 vs Deep Ripper:

Code: Select all

Score of LittleBeast 15 vs Deep Ripper: 9 - 5 - 86 [0.520]
...      LittleBeast 15 playing White: 7 - 0 - 43  [0.570] 50
...      LittleBeast 15 playing Black: 2 - 5 - 43  [0.470] 50
...      White vs Black: 12 - 2 - 86  [0.550] 100

Elo difference: 13.9 +/- 25.4, LOS: 85.7 %, DrawRatio: 86.0 %

SPRT: llr 0 (0.0%), lbound -inf, ubound inf
100 of 100 games finished.

Btw, here is another good example (in tie-break..) e.g
Just imagine...if we use to play small number of games,
Such as 100 games (per player)... then I mean simply that
The right Elo difference, standings etc. is mystery, right ?

As next Candidate match:
Eman 9.90 vs Polyfish 140124
Good Luck to Mr. Khalid Omar

Greetings

Posted: **Sat Jun 22, 2024 11:30 am**

UPDATE 3

Very good! Polyfish 140124 is next qualified player (by Khalid Omar)!

On the other hand,
Both eng versions are close in performance plus close in released dates too!
In other words, I mean simply that no BIG surprise at all....

Code: Select all

Score of Eman 9.90 vs Polyfish 140124: 31 - 34 - 435 [0.497]
...      Eman 9.90 playing White: 28 - 4 - 218  [0.548] 250
...      Eman 9.90 playing Black: 3 - 30 - 217  [0.446] 250
...      White vs Black: 58 - 7 - 435  [0.551] 500

Elo difference: -2.1 +/- 11.0, LOS: 35.5 %, DrawRatio: 87.0 %

SPRT: llr 0 (0.0%), lbound -inf, ubound inf
500 of 500 games finished.

Meanwhile (about the other rest engines),
That's all ..there will be no more Candidate tours for planning CS!
To be more clear, theirs latest Top engine vers. will be played out..

And please stay tuned,
Very soon the 1st official played results are coming...!

GOOD LUCK TO ALL AUTHORS !!

Best,
Sedat

Posted: **Sat Jun 22, 2024 12:25 pm**

UPDATE 4

And as I promised,
Here are the 1st played official results of SCCT NNUE CS LII

Very interesting.. for right now, we have 4 (four) Leading Engines:
Deep Blue, Polyfish, SF-PB and Stockfish - My Congratulations..!

Btw, to be honest I can't remember that after serious amount of games,
4 Top engines as Leaders..what about you? did you ever notice before ?)

LIVE RANKING VIA CUTECHESS (after 900+ games per Player):

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 SF-PB 120624                    5       8     909   50.7%   87.1% 
   2 D.Blue 180524                   5       8     909   50.7%   86.9% 
   3 Polyfish 140124                 5       8     910   50.7%   86.9% 
   4 Stockfish 150624                5       8     911   50.7%   87.7% 
   5 LittleBeast 15                  2       8     914   50.2%   88.0% 
   6 SF-PRO 160624                  -0       8     911   49.9%   87.2% 
   7 Corchess 180624                -2       8     911   49.7%   87.4% 
   8 Private 19Sf                   -2       8     909   49.7%   87.5% 
   9 Kookaburra 2                   -6       8     910   49.1%   87.4% 
  10 Killfish 150624               -10       8     912   48.6%   86.2% 

4553 of 7560 games finished.

Conditions:
2x Epyc 7B12, CuteChess, 1 Core, Ponder OFF, 30s+0.6s, Unique, 64 Hash, 4-MEN
Note: Sure where possible, each participant's move overhead is increased: 250
Be aware of that too as same words going during recent Candidate tours as well

Greetings )

Posted: **Sat Jun 22, 2024 1:06 pm**

UPDATE 5

Via this update, the standings slightly changed..especially the Top ones..
Great..for moment, Skynet's engine is current Leader - Congrats to Alex!

Code: Select all

Rank Name                          Elo     +/-   Games   Score    Draw 
   1 SF-PB 120624                    8       7    1337   51.2%   86.5% 
   2 D.Blue 180524                   3       7    1335   50.4%   86.7% 
   3 Stockfish 150624                2       6    1331   50.3%   88.6% 
   4 SF-PRO 160624                   1       7    1331   50.2%   86.3% 
   5 Corchess 180624                 1       7    1335   50.1%   86.9% 
   6 Private 19Sf                   -1       6    1336   49.9%   87.9% 
   7 Polyfish 140124                -1       7    1337   49.8%   86.9% 
   8 LittleBeast 15                 -2       7    1332   49.8%   87.4% 
   9 Kookaburra 2                   -5       7    1336   49.3%   86.8% 
  10 Killfish 150624                -7       7    1334   49.1%   87.2% 

6672 of 7560 games finished.

Note also that as next new update/s..
It's planning after Euro2024 match Georgia vs Czechia
Or maybe even after Turkey vs Portugal match...

Greetings )

Posted: **Sun Jun 23, 2024 5:10 am**

UPDATE 6

Well, here 2 (two) more entries: Brainlearn 28.1 and SFNNv6 16.1
Since in ongoing CS 'no size limit', so latest Brainlearn ver is in..

Code: Select all

Rank Name                         Elo     +/-   Games   Score    Draw 
   0 Brainlearn 28.1                1       8     861   50.1%   87.5% 
   1 LittleBeast 15                13      23      78   51.9%   91.0% 
   2 Corchess 180624               13      29      79   51.9%   86.1% 
   3 SF-PRO 160624                  9      31      76   51.3%   84.2% 
   4 Killfish 150624                9      29      80   51.2%   85.0% 
   5 D.Blue 180524                  9      24      80   51.2%   90.0% 
   6 Stockfish 150624               0      32      79   50.0%   82.3% 
   7 SF-PB 120624                  -9      21      78   48.7%   92.3% 
   8 SFNNv6 16.1                   -9      25      77   48.7%   89.6% 
   9 Private 19Sf                 -13      33      79   48.1%   81.0% 
  10 Polyfish 140124              -14      20      77   48.1%   93.5% 
  11 Kookaburra 2                 -18      27      78   47.4%   87.2% 

861 of 1848 games finished.

Meanwhile,
Very likely today, I hope to share the FINAL standings...
And any of the above player will contain min 1848 games
That means in the end...we will have serious data about
Which is Best of Best...sure I refer for under these cond...
Just in case of equal points by the highest ranked ones:
There will be Tie-Break duels for determining the Winner!

GOOD LUCK )

Posted: **Sun Jun 23, 2024 9:33 am**

Breaking News (23.06.2024):
1st of all, thanks to all authors..without you: these tours would not exist !!
Yes another new CS is organized...and as players of 3770+ participated..
What is new (via current tour)? only 1 (one) engine per author/s allowed
And instead of Balsa this time Unique openings are used..and as we see,
Any engine performance depends on our used cond. e.g via Balsa suite: LI
https://sites.google.com/site/computers ... -cs-51-2nd
If we are looking for eng improvements... the situation is 'piteous' for sure,
Where via Unique suite, it's clear that there is serious improvement, right?
And in same time that means via strong openings is 'worthless' to organize
3770+ tours, because the electricity is not cheap in my area, where I live..)
And without to not mention this I can not as well, some people prefer to say
Such as Perfect, Balsa openings are Drawish Btw, maybe Top engines are
Too Drawish? Who knows ?) However they suggest to be used other books
Suites..usually, they suggest the very weak suites..sure no problem..anyone
Is free to run, to use any opening, engine as he wish...But however, these
'Clever' experts are missing one small point, if via handicapped, weak suite:
1st They always forget to mention that these opening suites are very weak!
2nd for right, correct Elo measurements are required thousands of games
Per player..that's it..no more no less and 3rd I hope my postings will be useful..
On other hand, Skynet's engine is as 'BEAST' under these tournament cond!
And frankly soımetimes I wonder...(since SF-PB is based/derived from SF)
How can be possible that SF-PB is performing better than its mother (SF) ?
Sure I'm not a programmer, but looks like SF-PB is came from other planet..
And destroyed everything on field..reality is reality..!! Btw for anyone missed,
SF-PB has tiny size, capable to use BIN, in short, what we can need more?)
Where its real mother (SF) has no such great futures...sad...but very old SF
Vers. were capable to use BIN...(if I'm not wrong up to SF5 were capable...)
In other words, actually the wheel goes forward, not backward, but whatever..
And more strange, mostly Top ones which are considering as original (sure
I mean for NON-SF based engines), they are not capable to use BIN too ))
And my 6th strong sense says if SF will start to use BIN, they will start too )

The Champion: SF-PB 120624 - My Congratulations to Skynet, SF Team!
I congratulate the authors of D. Blue for 2nd place / SF-PRO for 3rd too!

For FINAL Standings:
https://sites.google.com/site/computers ... t-nn-cs-52

GAMES:
https://mega.nz/file/Kl4TBYiJ#dnKWq5KX8 ... mda4_LLDu0

Have a nice weekend )
Sedat

Posted: **Sun Jun 23, 2024 12:45 pm**

One thing more,
• Against time loses, any Engine's move overhead is increased to 250
Except SF-PRO, which comes without move overhead options, sad..
Very likely for these reasons, SF-PRO lost 3 (three) games on time (
I hope these engines which are not stable..will be fixed..if not simply
They will be out.. sure if future tours... thanks for your understanding,
Because 'Stability' is one of most important parts at least for my side!
In short, for fair..we have to fight against gaining very easy points...
Anyhow here is Screenshot, so you can check which are lost on time
E.g there are 12 games lost on time, mainly by Polyfish and SF-PRO:

Best,
Sedat

Posted: **Mon Jun 24, 2024 6:47 pm**

Hello Chess Friends,

As usually, I'm very pleased to announce that,
I plan to organize another exciting Duel match..!
But this time under more optimal tour conditions:
Blitz, MP, Ponder ON...

Some quick notes about the planning participants:
Both engines are Winners of recent tournaments: LI / LII
And it seems not bad idea to be in battle each other, right?

Btw,
Any guess: who will be the Grand Champion of this Duel?

Ok friends that's all for now, exception, soon as possible,
Minimum 300 games will be played out... Good Luck )

For more info (coming soon):
https://sites.google.com/site/computers ... n-vs-sf-pb

Best,
Sedat

Posted: **Tue Jun 25, 2024 7:42 pm**

UPDATE

Yes...SF-PB wins again..!! Congrats to Skynet and SF team!
I congratulate OrgZ too! No One can be always Number One!
Because any one may go down, but this is also true that as
Most important achievement is to get back up again, right?

On the other hand,
Once more we noticed that Skynet's SF-PB engine is very
Strong, powerful.. fully deserves to be called 'Born to Win'!
Meanwhile, (for anyone missed) according to overall stats:
SF-PB has many records as Number One..that means a lot!
Of course, as same words are going for OrgZ's engines too!

Be aware of that too,
In Duel matches: Champions titles are not given, as reasons:
Due to small contests (only 2 players, small number of games),
Anyhow I think these kind of Duels are worth to be organized,
Otherwise, if all tours were same, chess would not be boring?
Sure not good idea to run only Duels, because if running often:
It's not a very effective way to test all other Top chess engines,
But no doubt via Ratings (many players + games) provide much
More useful info about determining who is best, sure I respect
Other Testers as well, but not all of us have similar hardware,
Much free time etc. Btw, is that all enough ? not of course.. )

What I can say more,
Current Duel's database seems to be really not bad in quality!
Besides, there are not so many games out... I mean played
Under similar cond. such as on same, fast CPU machine etc.
But I admit too that the current draw percentage is high (95%),
As reasons: played MP, Blitz, Ponder ON..actually you know,
There are some ways to reduce the draws, e.g switching to
Much weaker openings, but just in case it's as lottery,mystery..
To see which one is stronger e.g in 300 games (per player),
Other reasonable alternatives are e.g we have to use 1 Core,
Bullet, Ponder OFF..or playing vs weaker opponents, then
The Draws are expecting to go down as well...or switching to
Directly to old, slow hardwares..then we'll see less Draws too

Ok...too many comments ) and here are the Duel's results:

Code: Select all

1   SF-PB 120624   +10/-4/=286 51.00%  153.0/300
2   C-Dragon v3.0  +4/-10/=286 49.00%  147.0/300

GAMES:
https://mega.nz/file/ykoygSTb#oMKfyrp09 ... tOTyCi7y6o

Best,
Sedat

Chess Forum

Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz

Re: Sedat Canbaz