There were some questions regarding of participating in the next SCCT Book Super League I
So i've decided to post here...
Some Notes about SCCT/Testo Leagues:
-The next SCCT Book Tournaments will played as Leagues
-The current qualified TOP 20 Books are the participants from SCCT IX Book Tournament
-Only the qualified TOP 20 Books will be participated in the next SCCT Super League I
-It's not allowed different other book participants in the next SCCT Super League I
-So if you wish to participate in SCCT Super League II,you should make a registration in Testo Book Second League I
-Only Testo Second League's Top 3 Books will qualify to play in the next SCCT Super League II
For more information:
I sent you an e-mail 3 days ago, please check your box and tell me what you think about this proposal. I will send you my updated book around 26/27 of november 2011.
Thanks...i have seen your mail
Later i will replay you
>In the Second league can participate with Polyglot book?
Thank you for your interest...
Not sure about Guido will allow testing polyglot books
But personally i think testing Polyglot books (under Fritz GUI) is not good idea
-Not very oftenly,but sometimes i've seen Houdini 2.0 crashes via Polyglot 1.4.67b
*Not sure about the latest release of Houdini 2.0c (regarding those crashes will appear or not)
-Polyglot books have big disadvantage:they can not play with Learning Functions ON
-In fast time controls,there is possibility the performance of Houdini engine to be effected by using Polyglot adapter
-SCCT Polyglot book participants will be tested in gauntlet mode
*Due to i am afraid to see crashes again by Houdini 2.0c via Polyglot adapter
As we know,in case of any polyglot crash...then Houdini and Polyglot exe fles are still in the system process
In other words,the exe files dont terminate...and that can down the performance of the rest book participants
And if those polyglot crashes will be appear again...then polyglot books will be not allowed in the next SCCT Book Tournaments
Honestly in my opinion,ctg book formats have more chances to be ranked at one of the TOP places in SCCT book competitions
Are you interested to be a TD in book competitions ?
If yes...its very simple,then you just need to open a registration table...
Note also that it will very hard... in case of testing many opening books in Auto232 mode
Or maybe you dont know how you can create book participants ?!
For example,here is a Tutorial about 'How to Create Book Participants':
I was wondering if you have gotten my email about my updated book?
Thank you for your interest...
Yes...i have seen your mail
Right now i am too busy,but later i will download your book
I have just to send you my updated book Mapanare 2.0a to compete in your first super league tournament. Please inform me if it is ok.
Please check your e-mail box
Yes...i downloaded it already and many thanks for your book update
I wish you good luck,
There were some hashtables discussions on RebornSE forum (and i've decided to post here too)
The discussions were about which hashtable size is better for 3 Minutes/Game on i7 980X machine:
512 mb or 128 mb hashtable size ?!
In my opinion, the best way of measuring the hashtable performance is to to test them against each other
So...currently i am testing both hashtables (128 MB against 512 MB)
Note:the Winner of the hashtable size will be used in the next SCCT Super League
Conditions of the hashtable match:
i7 980X @4.33GHz
Win XP Prof x64
Large Pages OFF
Hyper Threading OFF
Perfect 2012a book (alternate colors)
Engine:Houdini 2.0c x64 6 cores
And here are the current results afer 72 games
For right now,the elo performance:+29 elo in favor for 512 MB Hashtable size
1 512 MB Hashtable 39/72
2 128 Hashtable 33/72
-the both hashtables will be tested at least 100 games per player
-More games are needed,but anyway it seems 512 mb performs better than 128 mb (on i7 980X @ 4.33 GHz machine)
-In my previous testing (on my quads), 128 mb hashtable size was the Winner
1 512 MB Hashtable 75/148
2 128 Hashtable 73/148
For a serious statement (which hashtable is better ) is needed at least 1000 games,but probably i will stop the current match after 200 games
Still not sure about to change the hashtable conditions or not...
Actually the most important is what will be the opinion of SCCT Super League TOP 20's Book Makers
Maybe its better to create a poll...
> The Winner:512 MB Hashtable size:
That's the way it "should be"!
Since your games are around 3 sec/move and you probably get around 15000 kN/s on your i7 980X @4.33GHz with Houdini, you "fill" the Hashtable (in Houdini's case) with approx. 360 MBytes for each move, which is right for a 512 MB Hashtable size.
Note however that the kN/s you get would be slightly more with 128 MB Hashtable size.
But of course, since it is so small, in order to find the real gain of 512 MB Hashtable size vs. 128 MB Hashtable size, I am afraid you need much more than 1000 games!
Yes...agreed with some of your notes and even the current votes are showing that 512mb hashtable size is better
Actually 128 mb hashtable's performance (for blitz games) is better on my quads
As we see the current match results are still not very clear about which is better
In my opinion,the both hashtable size are very close in strength
And as i mentioned before,to be 100 % sure we need at least 1000 games per player
Maybe later (after SCCT Super League) i will resume the hashtable test and i will run min 1000 games
Houdini 2.0c x64 6 cores average kns values (on my machine- i7 980X @4.33GHz) are approx.18.000-19.000 kns,but not 15.000 kns !
Maybe without overclocking it will be around 15.000 kns
> Houdini 2.0c x64 6 cores average kns values (on my machine- i7 980X @4.33GHz) are approx.18.000-19.000 kns,but not 15.000 kns !
then in 3 sec you "fill" around 450 MB hash (instead of 360), which should make a 128 MB hashtable even more "insufficient"...
I am looking forward to your resuming your hashtable test and running min 1000 games!
> Actually 128 mb hashtable's performance (for blitz games) is better on my quads
If you follow my maths, with "slow" machines (5000 kN/s or less, on Houdini) that's what you get...
But in any case, the hashtable effect seems to be quite small, Elowise!
•Tournament Start Date:02 December 2011
SCCT Book Super League’s New Book Updates/Registrations (Total:19):
Opening Book Book Author Country
Attacker 6 Adriean Costinescu Romania
Blade 5.1 Jitan Stribal Japan
Bullseye 1.2 Clemens Keck Germany
FilthySanchez Mark Sabu USA
Guardian Angel 5 Giuseppe Petruccelli Italy
JapanTopTeam Marcos K. Matsuda Japan
Mapanare 2.0a Jean Disquin Venezuela
Muntaqim 3 Amir Zebedi UAE
Nano 2 Gabriel Nathans USA
PetaDeath Randy Sands USA
Pride 3 Alexssander Shirado Brasil
Prodigy 12 Maurice Chac Indonesia
Random 3 Tsvetan Milchev Bulgaria
Rompecoco 2c Girolamo Randazzo Venezuela
Sequoia Nick Carlin England
Sicilian Finale Munish Gupta India
The Winner Murat Guler Germany
TopTier 3 Audy Arandela USA
Uptown 3.1 BrandonSi USA
Note:Unfortunately no new book update only by Eduardo Koji
So,i will use the Hitman book,which has been played in SCCT IX
Fore More Information:
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill