Meanwhile,
In short: I can summarize about the attacks over my work etc.
As 'Sour Grapes' and I assume that many of you see the reality:
Someones can be so negative to best achievements etc. simply
Because they cannot have it themselves but this is a bad way!
What I can say more over, I suggest harder work, just because
It's hard, it does not mean to attack or to give up...try again..
Sooner or later the success will be in door..and do you see?
In end, who wins? as usually, best ones..proved many times!
Ok.. that's all for now.. and finely I feel much better !)
Greetings
Sedat Canbaz
-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
Re: Sedat Canbaz
Breaking News (08.09.2024):
1st of all, just don't blame and it is not any shame that Wang Hao is last...
Yes Wang is only one, which based just on human games, Winner: BN XI
And here are the both tours standings (merged NON-NN + NN into one...)
This time Unicarnio is said the last word!! where 2nd +3rd are DON vers
Goddess is 3rd rank too, 4th rank: Everest, OrionPax and Red Dragon!
In other words, (if we are going to check overall 2023 year performances):
Current produced tour results are not appeared due to luck!...sure I admit
Some books may suffer or just opposite..some may gain..sure not with all..
I mean the influences may be high with some books..as main reason, via
My creation and optimization methods, theirs weakest lines are becoming
Stronger..since I don't import/allow theirs previous played defeats/lost lines!
In same time that means, many of theirs lines are becoming more Drawish...
Anyhow just fun tour and once more my Congratulations to all rest Champs!
About next Number Ones 2024 CS: planning to be organized in Jan 2025
Otherwise, we may miss some newer Champs..since 2024 is still not over )
4rd Stage (2023) Champion: Unicarnio - Congratulations to Angel Morano!
FOR FINAL STANDINGS:
https://sites.google.com/site/computers ... -35-stage4
GAMES:
https://mega.nz/file/n0ZxUSZA#Bf4kDG3cU ... fTmpg1zuAY
Best,
Sedat
1st of all, just don't blame and it is not any shame that Wang Hao is last...
Yes Wang is only one, which based just on human games, Winner: BN XI
And here are the both tours standings (merged NON-NN + NN into one...)
This time Unicarnio is said the last word!! where 2nd +3rd are DON vers
Goddess is 3rd rank too, 4th rank: Everest, OrionPax and Red Dragon!
In other words, (if we are going to check overall 2023 year performances):
Current produced tour results are not appeared due to luck!...sure I admit
Some books may suffer or just opposite..some may gain..sure not with all..
I mean the influences may be high with some books..as main reason, via
My creation and optimization methods, theirs weakest lines are becoming
Stronger..since I don't import/allow theirs previous played defeats/lost lines!
In same time that means, many of theirs lines are becoming more Drawish...
Anyhow just fun tour and once more my Congratulations to all rest Champs!
About next Number Ones 2024 CS: planning to be organized in Jan 2025
Otherwise, we may miss some newer Champs..since 2024 is still not over )
4rd Stage (2023) Champion: Unicarnio - Congratulations to Angel Morano!
FOR FINAL STANDINGS:
https://sites.google.com/site/computers ... -35-stage4
GAMES:
https://mega.nz/file/n0ZxUSZA#Bf4kDG3cU ... fTmpg1zuAY
Best,
Sedat
-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
Re: Sedat Canbaz
Meanwhile,
Just I'd like to point also perhaps many of you aware too:
All of latest/recent XXXV Stages Winners (Champions) are
Belonging to openings played by Private books, where
The openings played by Public books are rated lower...
Actually, this is nothing new and I think it's quite normal that
Normally the Private openings have to perform better/stronger!
(Btw, as same words are going for all private engines as well )
Sure not always.. for example, as we know in other new XXXIII CS:
https://sites.google.com/site/computers ... -cs-53-2nd
The performance of overall Public books gained more points than
The Private ones..why ? simply because I think that many of the
Authors worked over several books... otherwise, if working only
Over one version, then I expect to see better points.. but no any
Problem.. since just fun tours ) but now, we learn/know more..!)
What I can I add more,
For experienced Top book makers, the available public opening
Lines are easier to be cracked..and in same time that means:
The available public openings are more potential to suffer..at
Least all the Number One Stage CS results indicate this, right?
Greetings )
Just I'd like to point also perhaps many of you aware too:
All of latest/recent XXXV Stages Winners (Champions) are
Belonging to openings played by Private books, where
The openings played by Public books are rated lower...
Actually, this is nothing new and I think it's quite normal that
Normally the Private openings have to perform better/stronger!
(Btw, as same words are going for all private engines as well )
Sure not always.. for example, as we know in other new XXXIII CS:
https://sites.google.com/site/computers ... -cs-53-2nd
The performance of overall Public books gained more points than
The Private ones..why ? simply because I think that many of the
Authors worked over several books... otherwise, if working only
Over one version, then I expect to see better points.. but no any
Problem.. since just fun tours ) but now, we learn/know more..!)
What I can I add more,
For experienced Top book makers, the available public opening
Lines are easier to be cracked..and in same time that means:
The available public openings are more potential to suffer..at
Least all the Number One Stage CS results indicate this, right?
Greetings )
-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
Re: Sedat Canbaz
Hello there )
1st of all,
Current ongoing tour is not a Championship, just in case..as before
I'll have serious difficulties to find stronger... sure I refer for
Close in dates/years openings...and this is not as joke...proved
Not only once, not twice...many times..!!
To be more clear, via this new tour, my target is simply that:
Recently (actually even in the past) these Number Ones are played..
But in recent Number Ones CS..as NN engine (for all Top books used):
Raubfish NNUE SC, I mean its strength is close to 3750 Elo, but where
In past, several Rybka eng vers used, which are close to 3000 Elo and
Now you can imagine.. On other hand, Rybka engines are not compatible
To use BIN books, that's why I realized to use Stockfish 5 (for all books)
Yes SF5 is stronger than Rybka, but I could find other any better solution..
E.g my goal is now to adapt more close conditions such as played in past..
Actually via Polyglot adapter is possible many of the engines to use BIN,
But no any idea what will be the influences.. e.g according to old tests,
Some of engines suffered via using external adapters..for these reasons:
I am not a very BIG fun of using external adapters!
In other words (this time via Stockfish 5, which capable to use BIN):
Let's see what will be the influences.. again from Past to Present..
As NNUE, we already know their performances..what about as NON-NN?
Can we say also that Perfect 16 will manage to be best? as before..
Who knows? maybe previous produced NN results were as luck etc. ?
And without to not mention this I can not as well,
Frankly, I disagree with those 'clever' guys who often prefer to say as
All books are randomly ranked, plus who say as dead drawish lines etc
Maybe theirs chess engines are dead, too drawish..and who knows?
Meanwhile, blaming the strongest opening theory is madness!! just saying!
But I admit that via NNUE 3700+ sure under optimal cond. is really boring!
Since the Top engines are playing/gaining with huge trained NN data...
But as reason here are not only openings...there are many other factors,
Which can play as serious influence to appear more draws for examples
Some of the Top engines (even as NON-NN) are more drawish..proved already!
E.g Raubfish SC can be as reference...in that case....moreover Hardware speed,
Time Controls, MP, Ponder ON etc. are playing as an important role! and
That's not all, what about NN data? via BIG NN asistane much more draws!
Do you see? the openings are not only the reason....Btw, can you create
A strong engine 3700 + to be not so Drawish? and to be not close to SF ?)
And if asking about my work..via my Book + Eng tours I've done my best..!!
And exc. my organized tours over long past years..I'm author of many books,
Sure I am creator of many other opening suites: Balsa, Champ, Unique etc.
And mainly for these kind of reasons, in case of NN 3700+ tours:
I use mainly Bullet, 1 Core, Balsa, Unique etc sure mainly for fun, if not:
Usually I say: what's point of using BIG umbrella, if no much rain on horizon ?)
And in our hobby, there are many who prefer testing various cond. my respect!
Yes, we are all different, if we were all same, World would NOT be too boring ?)
Oh, sorry my friends,..I almost forgot.. here are current LIVE rankings:
Sure it's too early for any conclusions.. since we need much more games..
But anyhow, Perfect is Perfect...right ?
Conditions:
2x Epyc 7B12, CuteChess, 1 Core, Ponder OFF, 3m+05s, 128 Hash, 4-MEN
Note also that all Number Ones are playing via Stockfish 5 (3300+ Elo)
And please stay tuned.. more updates soon as possible..
Greetings
1st of all,
Current ongoing tour is not a Championship, just in case..as before
I'll have serious difficulties to find stronger... sure I refer for
Close in dates/years openings...and this is not as joke...proved
Not only once, not twice...many times..!!
To be more clear, via this new tour, my target is simply that:
Recently (actually even in the past) these Number Ones are played..
But in recent Number Ones CS..as NN engine (for all Top books used):
Raubfish NNUE SC, I mean its strength is close to 3750 Elo, but where
In past, several Rybka eng vers used, which are close to 3000 Elo and
Now you can imagine.. On other hand, Rybka engines are not compatible
To use BIN books, that's why I realized to use Stockfish 5 (for all books)
Yes SF5 is stronger than Rybka, but I could find other any better solution..
E.g my goal is now to adapt more close conditions such as played in past..
Actually via Polyglot adapter is possible many of the engines to use BIN,
But no any idea what will be the influences.. e.g according to old tests,
Some of engines suffered via using external adapters..for these reasons:
I am not a very BIG fun of using external adapters!
In other words (this time via Stockfish 5, which capable to use BIN):
Let's see what will be the influences.. again from Past to Present..
As NNUE, we already know their performances..what about as NON-NN?
Can we say also that Perfect 16 will manage to be best? as before..
Who knows? maybe previous produced NN results were as luck etc. ?
And without to not mention this I can not as well,
Frankly, I disagree with those 'clever' guys who often prefer to say as
All books are randomly ranked, plus who say as dead drawish lines etc
Maybe theirs chess engines are dead, too drawish..and who knows?
Meanwhile, blaming the strongest opening theory is madness!! just saying!
But I admit that via NNUE 3700+ sure under optimal cond. is really boring!
Since the Top engines are playing/gaining with huge trained NN data...
But as reason here are not only openings...there are many other factors,
Which can play as serious influence to appear more draws for examples
Some of the Top engines (even as NON-NN) are more drawish..proved already!
E.g Raubfish SC can be as reference...in that case....moreover Hardware speed,
Time Controls, MP, Ponder ON etc. are playing as an important role! and
That's not all, what about NN data? via BIG NN asistane much more draws!
Do you see? the openings are not only the reason....Btw, can you create
A strong engine 3700 + to be not so Drawish? and to be not close to SF ?)
And if asking about my work..via my Book + Eng tours I've done my best..!!
And exc. my organized tours over long past years..I'm author of many books,
Sure I am creator of many other opening suites: Balsa, Champ, Unique etc.
And mainly for these kind of reasons, in case of NN 3700+ tours:
I use mainly Bullet, 1 Core, Balsa, Unique etc sure mainly for fun, if not:
Usually I say: what's point of using BIG umbrella, if no much rain on horizon ?)
And in our hobby, there are many who prefer testing various cond. my respect!
Yes, we are all different, if we were all same, World would NOT be too boring ?)
Oh, sorry my friends,..I almost forgot.. here are current LIVE rankings:
Sure it's too early for any conclusions.. since we need much more games..
But anyhow, Perfect is Perfect...right ?
Code: Select all
Rank Name Elo +/- Games Score Draw
1 Perfect 16 57 33 130 58.1% 68.5%
2 Ultra-Blitz 32 30 132 54.5% 74.2%
3 Xmas2640 19 29 131 52.7% 76.3%
4 Experimental 16 31 130 52.3% 73.8%
5 My Book 13 31 130 51.9% 73.1%
6 Rybka III 11 29 127 51.6% 76.4%
7 Magnificent 8 32 133 51.1% 70.7%
8 Hurricane -5 31 132 49.2% 72.7%
9 Yograj -5 36 130 49.2% 63.1%
10 Perfect 13x -8 33 127 48.8% 70.9%
11 CompMaster -13 34 130 48.1% 68.5%
12 OmPrakash -14 35 128 48.0% 66.4%
13 Perfect 13 -21 31 131 46.9% 72.5%
14 Rybka II -89 38 131 37.4% 58.0%
911 of 4550 games finished.2x Epyc 7B12, CuteChess, 1 Core, Ponder OFF, 3m+05s, 128 Hash, 4-MEN
Note also that all Number Ones are playing via Stockfish 5 (3300+ Elo)
And please stay tuned.. more updates soon as possible..
Greetings
-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
Re: Sedat Canbaz
UPDATE
Well, for right now (the competition with 2006-2010 opening theory):
Perfect 16 is leading as before...with 57% win. perc.
Where P16 opening book's draw ratio is close to 65%
In other words, again we see as 'unbelievable' record! but we will see..since
The tour is not over..Note also min. 650 games (per player) will be played...
On the other hand,
Under these conditions, which is more adapted to past..
Rybka II (the oldest - 2006) is currently as last rank..
I mean e.g in the recent official Number Ones NN CS,
Rybka II book is performed much better...so now is
More clear that NN influence is played as serious role!
Sure not for all of course..but for some it seems so...
Well, for right now (the competition with 2006-2010 opening theory):
Perfect 16 is leading as before...with 57% win. perc.
Where P16 opening book's draw ratio is close to 65%
In other words, again we see as 'unbelievable' record! but we will see..since
The tour is not over..Note also min. 650 games (per player) will be played...
On the other hand,
Under these conditions, which is more adapted to past..
Rybka II (the oldest - 2006) is currently as last rank..
I mean e.g in the recent official Number Ones NN CS,
Rybka II book is performed much better...so now is
More clear that NN influence is played as serious role!
Sure not for all of course..but for some it seems so...
Code: Select all
Rank Name Elo +/- Games Score Draw
1 Perfect 16 50 23 310 57.1% 65.2%
2 My Book 19 21 313 52.7% 69.0%
3 Experimental 15 21 308 52.1% 71.8%
4 Rybka III 9 20 312 51.3% 71.8%
5 Ultra-Blitz 8 21 313 51.1% 69.0%
6 Magnificent 2 22 310 50.3% 69.0%
7 Hurricane 0 20 311 50.0% 73.0%
8 Perfect 13x -3 21 309 49.5% 69.9%
9 Yograj -6 23 312 49.2% 63.8%
10 CompMaster -6 20 310 49.2% 71.9%
11 Xmas2640 -9 19 310 48.7% 75.5%
12 OmPrakash -11 22 310 48.4% 67.7%
13 Perfect 13 -18 20 311 47.4% 71.7%
14 Rybka II -49 23 311 42.9% 63.3%
2175 of 4550 games finished.-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
Re: Sedat Canbaz
Breaking News (11.09.2024):
1st of all and as I stated earlier: this is not a CS...just organized to check
NON-NN influences, because this time Stockfish 5 is used for all players..
For example, the previous tour played via Raubfish NN..now as NON-NN
And the present tour is adapted more close conditions as played in past!
Moreover, my target is not to prove that I may create not so weak books)
Many of you know Me over long past years plus nowadays we have great
Talents, who create superior powerful books too...Note also that I did not
Touch/tune (by hand) any of original drawn+won openings they are left as
They have to be..but till 2009, I was quite active over newer/stronger move
Ideas that never played before..actually as reference, you can check also
All individual stats etc, where you'll notice also that P16 lost to some strong
Champs..but in past, I'd not allow (before final release) to public ones to win)
At least to available opening lines)
But anyhow, I'm satisfied..if not, I have to be ungrateful.. because in almost
All tours (sure in my tour categories) as usual my work is not so bad, right?
Btw, for comparing (NN vs NON-NN).. here are the recent NN tour: Results
So now we've more data about NN influences, for some better / worse etc.
But this is also true, current standings are more close to the results of past..
Ok dear chess friends that's all for now...and many thanks for your interest )
The Leader (2006-2010) Perfect 16 - Exc. Me) Congrats to all Authors too!
FOR FINAL STANDINGS:
https://sites.google.com/site/computers ... -2006-2010
GAMES:
https://mega.nz/file/v1IHUCKL#s0l2Byf-O ... VOhb-wJX_E
Best,
Sedat
1st of all and as I stated earlier: this is not a CS...just organized to check
NON-NN influences, because this time Stockfish 5 is used for all players..
For example, the previous tour played via Raubfish NN..now as NON-NN
And the present tour is adapted more close conditions as played in past!
Moreover, my target is not to prove that I may create not so weak books)
Many of you know Me over long past years plus nowadays we have great
Talents, who create superior powerful books too...Note also that I did not
Touch/tune (by hand) any of original drawn+won openings they are left as
They have to be..but till 2009, I was quite active over newer/stronger move
Ideas that never played before..actually as reference, you can check also
All individual stats etc, where you'll notice also that P16 lost to some strong
Champs..but in past, I'd not allow (before final release) to public ones to win)
At least to available opening lines)
But anyhow, I'm satisfied..if not, I have to be ungrateful.. because in almost
All tours (sure in my tour categories) as usual my work is not so bad, right?
Btw, for comparing (NN vs NON-NN).. here are the recent NN tour: Results
So now we've more data about NN influences, for some better / worse etc.
But this is also true, current standings are more close to the results of past..
Ok dear chess friends that's all for now...and many thanks for your interest )
The Leader (2006-2010) Perfect 16 - Exc. Me) Congrats to all Authors too!
FOR FINAL STANDINGS:
https://sites.google.com/site/computers ... -2006-2010
GAMES:
https://mega.nz/file/v1IHUCKL#s0l2Byf-O ... VOhb-wJX_E
Best,
Sedat
-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
Re: Sedat Canbaz
UPDATE
1st of all, today let Me start please, with this new question:
Why almost all Top NN 3700+ programmers are unsuccessful,
Fail to improve their engines with strong short openings?
Let's say Balsa v500 (up to 5 moves), do you believe that
Balsa is dead, it's too drawish opening suite etc. ? If so..
I hardly doubt..but just in case..you are a pawn that
You do not know that it's a piece of chess game!
Frankly I see this situation as 'Comedy of Errors' / 'Sour Grapes'!!
Yes..some guys, who are not capable how to improve theirs eng
Strength etc. unfortunately they prefer to say for our strong short
Openings such as dead, drawish etc. hehe)..anything more else ?)
Perhaps the 'golden' key in the hand of AlphaZero team..ask them..
Who knows? they introduced the 1st NN.. to our chess community..
Via beating SF8 with clear margin..Btw, sure this is other question,
But their organized match was unfair such as duel: NON-NN vs NN
Continuing ...
Again who knows? maybe AlphaZero may help here how to improve?
Because since a long time, 3700+ Elo improvement is almost dead..
Sure here again I refer mainly for short (not deep) strong openings!
Btw, just a test via deep strong openings: situation is more critical,
Much less Elo improvements are expecting for sure...!
What I can say more,
Be sure..the strong opening theory is not dead at all or not so drawish..
At least under SCCT.. e.g I run many tours to determine which are best,
Strongest books as well and not since yesterday...since many years!
In other words, I can summarize also all over these issues such as,
It is time for a new revolution..to learn, to originate better ideas!
But just do not blame the strong opening lines.. and hope helps!
Have a nice day )
Sedat
1st of all, today let Me start please, with this new question:
Why almost all Top NN 3700+ programmers are unsuccessful,
Fail to improve their engines with strong short openings?
Let's say Balsa v500 (up to 5 moves), do you believe that
Balsa is dead, it's too drawish opening suite etc. ? If so..
I hardly doubt..but just in case..you are a pawn that
You do not know that it's a piece of chess game!
Frankly I see this situation as 'Comedy of Errors' / 'Sour Grapes'!!
Yes..some guys, who are not capable how to improve theirs eng
Strength etc. unfortunately they prefer to say for our strong short
Openings such as dead, drawish etc. hehe)..anything more else ?)
Perhaps the 'golden' key in the hand of AlphaZero team..ask them..
Who knows? they introduced the 1st NN.. to our chess community..
Via beating SF8 with clear margin..Btw, sure this is other question,
But their organized match was unfair such as duel: NON-NN vs NN
Continuing ...
Again who knows? maybe AlphaZero may help here how to improve?
Because since a long time, 3700+ Elo improvement is almost dead..
Sure here again I refer mainly for short (not deep) strong openings!
Btw, just a test via deep strong openings: situation is more critical,
Much less Elo improvements are expecting for sure...!
What I can say more,
Be sure..the strong opening theory is not dead at all or not so drawish..
At least under SCCT.. e.g I run many tours to determine which are best,
Strongest books as well and not since yesterday...since many years!
In other words, I can summarize also all over these issues such as,
It is time for a new revolution..to learn, to originate better ideas!
But just do not blame the strong opening lines.. and hope helps!
Have a nice day )
Sedat
-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
Re: Sedat Canbaz
UPDATE 2
Well, I just realized to test latest Stockfish 17,
Which is super strong close to 3800+ plus free !
That means a lot really...thanks a lot to SF team!
And here are latest new results, enjoy
Bullet (30s+0.6s): + 6 Elo diff. (in favor for SF17)
-----------------------------------------------------
Blitz (2m+1s): + 1 Elo diff. (in favor for SF17)
GAMES:
https://mega.nz/file/3hhmmBiS#YpCICwKMH ... WAz8k1gzNg
Conditions / Some Notes:
2x Epyc 7B12, CuteChess 1.3.1, Ponder OFF, 1 Core, 4-MEN, Balsa v500,
Note (for fair cond.): all openings are repeated with switched colors..
Note also that both Top SF engines are played via move overhead: 400
But anyhow, I have bad news too, reality is reality..
E.g latest Stockfish 17 is not so stable...sad indeed..
E.g SF17 is lost 1 (one) game on time (via Bullet)
On other hand, we are in 2024..but still we see
Instability, bad issues...moreover, SF team is based
On army of Top authors...but unfortunately non of them
Can improve SF17 in better way, sure I'm not referring
About its strength..I mean simply about eng stability!
This is a very important part.. at least for my side!!
By the way,
Soon as possible, I hope to share other new NON-NN tests...
Duel match: Stockfish 10 vs Stockfish 11.. we will see
What is Elo improvements and we will see their stability too..
And the test is exactly under same Bullet cond. via Balsa v500 too!
And let's see NON-NN Top engines will be too drawish or not ?!
Because in case of 3800 Elo vs 3800 Elo is just waste of time..
E.g as we see, no much Elo improvement..in more than 6 months..
Maybe SF team Is too busy with handicapped (weak) opening lines?
BTW, just don't say/suggest to use these how do they say 'unbalanced'
Lines.. because just in case...if my car has 'unbalanced'
Vehicle problems...my 1st job to be in car repair service !)
Sorry here...and it seems, via NN 3800+ harder to add NN Elo points...
Ok.. I already forgot SF Elo improvements...and let's hope also next
Stockfish 18 to be stable as before (like in the past) !!
In other words,
I hope all SF Authors will pay more attention about Eng stability...
Sure I refer here before final official releases... thanks in advance
And please stay tuned... very soon SF10 vs SF11 is coming!
Greetings )
Well, I just realized to test latest Stockfish 17,
Which is super strong close to 3800+ plus free !
That means a lot really...thanks a lot to SF team!
And here are latest new results, enjoy
Bullet (30s+0.6s): + 6 Elo diff. (in favor for SF17)
Code: Select all
1 Stockfish 17 +28/-12/=960 50.80% 508.0/1000
2 Stockfish 16.1 +12/-28/=960 49.20% 492.0/1000
Note: 64 MB Hash is used for Bullet / DrawRatio: 96%
Blitz (2m+1s): + 1 Elo diff. (in favor for SF17)
Code: Select all
1 Stockfish 17 +4/-2/=494 50.20% 251.0/500
2 Stockfish 16.1 +2/-4/=494 49.80% 249.0/500
Note: 128 MB Hash is used for Blitz / DrawRatio: 99%
GAMES:
https://mega.nz/file/3hhmmBiS#YpCICwKMH ... WAz8k1gzNg
Conditions / Some Notes:
2x Epyc 7B12, CuteChess 1.3.1, Ponder OFF, 1 Core, 4-MEN, Balsa v500,
Note (for fair cond.): all openings are repeated with switched colors..
Note also that both Top SF engines are played via move overhead: 400
But anyhow, I have bad news too, reality is reality..
E.g latest Stockfish 17 is not so stable...sad indeed..
E.g SF17 is lost 1 (one) game on time (via Bullet)
On other hand, we are in 2024..but still we see
Instability, bad issues...moreover, SF team is based
On army of Top authors...but unfortunately non of them
Can improve SF17 in better way, sure I'm not referring
About its strength..I mean simply about eng stability!
This is a very important part.. at least for my side!!
By the way,
Soon as possible, I hope to share other new NON-NN tests...
Duel match: Stockfish 10 vs Stockfish 11.. we will see
What is Elo improvements and we will see their stability too..
And the test is exactly under same Bullet cond. via Balsa v500 too!
And let's see NON-NN Top engines will be too drawish or not ?!
Because in case of 3800 Elo vs 3800 Elo is just waste of time..
E.g as we see, no much Elo improvement..in more than 6 months..
Maybe SF team Is too busy with handicapped (weak) opening lines?
BTW, just don't say/suggest to use these how do they say 'unbalanced'
Lines.. because just in case...if my car has 'unbalanced'
Vehicle problems...my 1st job to be in car repair service !)
Sorry here...and it seems, via NN 3800+ harder to add NN Elo points...
Ok.. I already forgot SF Elo improvements...and let's hope also next
Stockfish 18 to be stable as before (like in the past) !!
In other words,
I hope all SF Authors will pay more attention about Eng stability...
Sure I refer here before final official releases... thanks in advance
And please stay tuned... very soon SF10 vs SF11 is coming!
Greetings )
-
Sedat Canbaz
- Posts: 569
- Joined: Mon Jun 19, 2023 8:49 am
- Has thanked: 662 times
- Been thanked: 1970 times
Re: Sedat Canbaz
UPDATE 3
Bullet NON-NN: + 54 Elo diff. (in favor for SF11)
And via NON-NN: no any game is recorded to be lost on time !!
In other words, it seems all very stable... great news for sure !
Conditions / Some Notes:
2x Epyc 7B12, CuteChess 1.3.1, Ponder OFF, 1 Core, 4-MEN, Balsa v500,
Note (for fair cond.): all openings are repeated with switched colors
Both SF engines are played with Move overhead: 200
As FINAL Conclusion,
Classic is Classic...it is better, more fun and very stable!
Sure NON-NN is weaker...but stability is more important!
If nothing else, if comparing the latest SF17 release....
Meanwhile, this is nothing new.. e.g with my older hardware,
During the famous Duel Match: Book Vs NNUE III:
https://sites.google.com/site/computers ... nn-vs-book
Again and again the older SF NN (including SugaR) are lost on time too!
What is changed ? not so much...same story as in past..sad, but true!
Btw, what does it mean 'Classic' word: 'High quality'
At least under these played conditions, it seems to be so!
As we see..there are better stable games plus the matches are
Played via strong opening lines... besides, both SF vers produced:
69% draws !! you know.. not so high numbers, right?
Because the present latest test was with Top engines 3500 -3600 Elo points...
Plus, under these cond. more than 50+Elo improvement (between both vers.)
And frankly.. as a TD / Tester: I hope and I love to see similar good news...
Otherwise, just in case newer releases...sure with time forfeits etc. then
I feel like there is something wrong with my hardware/tournament setup !)
And I hope also that,
All my produced data to be useful for programmers ...if not..will be sad..
But I can live with that.. !)
Ok dear friends...thanks for reading, interest..see you later )
Best,
Sedat
Bullet NON-NN: + 54 Elo diff. (in favor for SF11)
And via NON-NN: no any game is recorded to be lost on time !!
In other words, it seems all very stable... great news for sure !
Code: Select all
1 Stockfish 11 +230/-75/=695 57.75% 577.5/1000
2 Stockfish 10 +75/-230/=695 42.25% 422.5/1000
Note: 64 MB Hash is used for Bullet / DrawRatio: 69%
2x Epyc 7B12, CuteChess 1.3.1, Ponder OFF, 1 Core, 4-MEN, Balsa v500,
Note (for fair cond.): all openings are repeated with switched colors
Both SF engines are played with Move overhead: 200
As FINAL Conclusion,
Classic is Classic...it is better, more fun and very stable!
Sure NON-NN is weaker...but stability is more important!
If nothing else, if comparing the latest SF17 release....
Meanwhile, this is nothing new.. e.g with my older hardware,
During the famous Duel Match: Book Vs NNUE III:
https://sites.google.com/site/computers ... nn-vs-book
Again and again the older SF NN (including SugaR) are lost on time too!
What is changed ? not so much...same story as in past..sad, but true!
Btw, what does it mean 'Classic' word: 'High quality'
At least under these played conditions, it seems to be so!
As we see..there are better stable games plus the matches are
Played via strong opening lines... besides, both SF vers produced:
69% draws !! you know.. not so high numbers, right?
Because the present latest test was with Top engines 3500 -3600 Elo points...
Plus, under these cond. more than 50+Elo improvement (between both vers.)
And frankly.. as a TD / Tester: I hope and I love to see similar good news...
Otherwise, just in case newer releases...sure with time forfeits etc. then
I feel like there is something wrong with my hardware/tournament setup !)
And I hope also that,
All my produced data to be useful for programmers ...if not..will be sad..
But I can live with that.. !)
Ok dear friends...thanks for reading, interest..see you later )
Best,
Sedat