Chess Forum

Hello Chess Friends,

I'm pleased to announce,
Here is my latest tutorial:



For anyone missed, interested for previous part 1:
https://open-chess.org/viewtopic.php?f= ... &start=100


Best,
Sedat

Hello Chess Friends,

I'm pleased to announce, new Duel is started:
SF-PB 160823 avx2 Vs CorChess 140823 avx2

And so far (after 173 out of 310 games):
Current score: +10 Elo (in favor for SF-PB)

Note that current test idea is born after checking
Oliprog older test, where CorChess managed to win:
viewtopic.php?t=140

Sure I've done my best to adapt to Oliprog testing..
And I can't say that our both used cond. are same!
At least I can say, now I test newer eng versions!

But anyhow, I think that sometimes...
It's worth giving a try )... more better, right ?)
Plus now we'll have new data for comparing!

By the way, SCCT conditions:
2x EPYC 7B12, Windows 10 x64, HT + LP Disabled, Cutechess,
1 Core, Ponder ON, 2m+1s, Unique 170623,128 MB hash, 4-MEN

Oliprog conditions:
AMD A10-9700, Windows 10 x64, HT + LP Enabled, Fritz GUI,
2 Threads, Ponder ON, 4m+1s, Unique 170623, 1 GB hash, 5-MEN

In other words,
Our both used cond. differences are more...but
Same of our used conditions are exactly same!

More info soon and please stay tuned )

Best,
Sedat

SCCT Duel - CorChess vs SF-PB is ended:
About +7 Elo (in favor for SF-PB 160823)

The Winner: SF PB - Congrats to Skynet, SF team!
The Overall Draw-Ratio (based on 310 games): 95%

GAMES:
https://mega.nz/file/OpxS1a4A#T_CAhwxYR ... SHNpWoz6Zs

More Details,
It's clear that the overall draw ratio is increased!
Actually nothing strange...reason, SF team seems to
Be working on optimizing over Evalfiles, plus since
Unique 170623 suite is publicly available so I assume
That is easier vs current opening lines (to be tuned)
E.g as reference/evidence, when current openings were
As private..then the overall draws were about 85%-90%
E.g now, as we see the overall draws are close to 95%
I mean, Unique openings are became drawish as Balsa!
Btw, I have to keep next newer Unique ver as private)
If nothing else,
We'll see more fun for sure...at least for my side!

As final words, it's clear also that:
Mostly of SF-PB vers perform not bad via SCCT conditions!)


Best,
Sedat

Meanwhile (for anyone missed, interested...)
I think that current useful Hash test is missing here...
https://open-chess.org/viewtopic.php?f= ... 0&start=30

Hello there )

1st of all,
This test is done due to some people have no clue
What is going on...and really sorry to say that,
For example, like Deeds member, for more details:
https://open-chess.org/viewtopic.php?f=4&t=3618

SCCT Hash-Table Test: 128 MB Vs 512 MB

Elo difference: 6 (in favor for 128 MB)
Games:
https://mega.nz/file/XlI2lAza#842Rgurpn ... vfD4Epfwv0

Draw-Ratio (based on 400 games) is not so high 90%

Conditions:
2x Epyc 7B12, CuteChess, Ponder OFF, 1 Core, Unique Book, 1m+1s, 4-MEN
Note that all matches are played via GUI's concurrent option: 64
And for each player is installed another folder copy of RaptoRa engine
I mean that both current players are used to play from different eng exe

Note also that:
These results will be published later on SedatChess site, because
Working on two sites is not so easy as it looks.. thanks for understanding

Greetings

And just another of my recent postings (posted on OpenChess)

Hello Friends,

Meanwhile (for anyone missed, interested...):
Here are some info, which can be as useful tips
To see about which test conditions are best...

1st of all,
It's not bad idea to run own strength testings!
In this way, you will have a chance to see....

Sure and as we know (since 2020, NNUE on scene),
The situation is not same as before...but however,
Running some serious pre-strength test would be not
So bad idea..e.g Draw, HT, Hash, Learning etc testings!
Even some compile strength testings you may run, e.g
Bmi2 or avx2 or avx or popcnt.. may perform better!
Or running a Draw test, sure depending on cond. but
Some NNUE engines may produce less draws..that
Means in same time.via best optimal cond., you may
Get higher points..or just opposite..for this reason:
Once more I say, please run your own tests,
Sure I am referring before playing serious tours!

Btw, as I stated earlier:
In case of using Large pages ON, be careful please!!
Sure it all depends, but if we are running long-matches...
We can not expect so much accurate results, for more info:
https://learn.microsoft.com/en-us/windo ... ge-support

By the way,
It's my pleasure to share my experience with all of you!
E.g when I was as online player (on Playchess server):

1. Hyper-Threading (SMT) was as disabled on my hardwares!
As reason, I've noticed that Rybka is performed better..!
Otherwise, not only Rybka, plus some other engines suffered,
I mean if running Ponder ON via Hyper-Threading as enabled..
2. Large Pages were as disabled on my operating systems!
3. My used Rybka engine vers were played via 128 MB hash

Exception the above important influences, you know that
For better performance, it's required also fast hardware!
Plus super strong book too, but via frequent book updates!
Otherwise, I think that no chances to be one of Top ranks!

What I can say more (about during 2008-2009 years),
Frankly, before entering to be as online player etc.
I've played many serious book strength testings!plus
I created newer strong lines that never played before!
And I noticed later (during online) the opponents were
Out of earlier of books, where my book was continued
To play in deeper opening lines...sure as a result,
In most cases, these new lines played as main role to
Get better points...

Sure there many other techniques...but that's all for now!
And I hope also that my current experience will be helpful!

Greetings )

New: SCCT - Top Similarity Overview - 28.08.2023

Hello there )

As usual, I'm pleased to announce that,
Here are latest Top Similarity Results!

And to be more clear,
Main target of running current sim test:
Soon a new SCCT rating is coming...and
As far as possible, I plan to test engines,
Which are less than 75% in similarity!
Thanks for your understanding!

More Details,
So far new 22 SF based engines are tested..
Where mostly of them are not very close !!
But several are almost identical in playing
Styles..and as reason based on same source!

Btw, here are those 10 engines, which have
The highest similarities (each other) so far:
Download all the rest similarity results:
https://mega.nz/file/Tw4H0a4I#10oaFgM9I ... bxxKPehic4

Test Conditions:
Ryzen7 4800H, Simex2, 500 Openings, 1 Thread, Time: 100 ms, 64 MB Hash
Note: Polyfish 210623 and SF-PB 190623 tested again..to see the influence
With new conditions..so it is clear that even 500 openings are enough!

As Final Words,
I hope to see newer and better original ideas! otherwise if all
Engines will be based on same code..then what is the point ?)
I understand of course that the target is to use BIN books..but
Too many cooks spoil the stew )


Best,
Sedat

New Update:
CatroPOLY 310723 and Cool-Iris 11.10 are entered...

By the way,
Here are the chess engines, which are not so close:
Note: Lower percentage values are better!

And how nice to see such Top chess engines,
Which are based on more original own ideas!

What I can say more... and again and again,
RapTora is RapTora: most different than all!
Not via papers of course, but in real facts!
Same words going for many others as well!


Best,
Sedat

By the way,
When I run/test something new: no peace for my side...
Especially when I'm very active on computer chess forum/s...
Actually I can..but I've no assistant.. so no free time for all!
Plus in this way, I think that the MODs job will be easier!

And mainly for these kind of reasons, since today:
I'll try to stay as not so frequent member at any forum!
Plus...I do not want to make any double standard...

Sure in next September,
You can follow my all chess activities from SedatChess site!

Thank you for your understanding and reading!

Best,
Sedat

Hello Chess Friends,

Just I'd like to inform you that,
The Similarity page is created, sure many more engines entered!

For More Details about Simex 2 - Top Similarity Results III:
https://sites.google.com/site/computers ... x-aug-2023


Best,
Sedat

Meanwhile (about the planning new rating),

I hope to test:
- Up to 5 (five) engine versions by known Author/s..
- Up to 75% (in engine similarity, sure each other...)
- Mainly new eng versions (never tested..) and 3700+
- There will be qualification leagues... for examples:
Second League tour: Bullet 1 Core, Ponder OFF etc.
Super League: Blitz, MP, Ponder ON etc sure and as
Usual these tours will be played under fast machine:
2x [Dual CPU] AMD EPYC 7B12 128 Cores/256 Threads

Greetings