Naum 3.1

[Ampzilla]

Naum 3.1 was sent via email to me 3:30PM

[Marek Soszynski]

From the Naum website http://www.geocities.com/naum_chess/

Naum 3.1 is currently being distributed for free to those who purchased Naum 3 !
(If you don't receive your free update by May,5th please email the customer support)

If you want to purchase Naum 3.1 the prices are as follows:

* Single processor version
o Customers who purchased Naum 2.2: 22 EUR
o New customers: 30 EUR

* Multiple processor version
o Customers who purchased Naum 2.2: 45 EUR
o New customers: 60 EUR

[Graham Banks]

The readme file:

Naum 3.1 is distributed in order to fix a couple of bugs found. It should also be stronger then Naum 3.

Main changes compared to Naum 3:
- Fixed EGTB initialization bugs for the UCI mode.
For instance, setting EGTB cache size to anything above 64MB was causing EGTB cache size to be reduced to 32MB.
There were also some instances where Naum would not update EGTB settings after they were changed by user in the UCI parameters window.

- There are 2 new parameters that influence how search pruning is done:
'pruning level' - This parameter can be very useful for the analysis. Setting it to a negative value will make Naum prune less, slowing the search, but increasing its quality. Setting it to e.g. -100 will frequently cause Naum to find wins 2-3 plies earlier. If you want to use it for tuning, then you should try smaller values (e.g. -5 or 5).
'pruning type' - You can choose 'normal' or 'smart' pruning algorithm. It seems that smart algorithm gives better results against most of the opponents, but the 'normal' type may be better for your conditions.

- Many users complained that every time they start Naum MP, the GUI resets number of threads to 1 (default value).
In order to fix this, I decided to read default values for UCI parameters from the naum.cfg.
Old version was ignoring naum.cfg in the UCI mode, but now it will use it to set default values for the UCI parameters.
So if you want Naum to use 2 threads by default, you should create naum.cfg file in the same folder as your naum.exe and put MAX_THREADS = 2 in it.
For instance, if you use Naum for analysis and you want to have 'preserve hash' set, use 4 threads, and have 'pruning level' set to -10 without having to set these parameters every time you start your GUI, you should add naum.cfg file with the following lines in it:
PRESERVE_HASH = 1
MAX_THREADS = 4
PRUNING_LEVEL = -10

- New version has changes that should increase ELO, but I am not going to estimate by how much.

[Tony Thomas]

I havent received mine yet, hopefully its worth the wait..

Tony

[geots]

I havent received mine yet, hopefully its worth the wait..

Tony

Alex told me all 3.0 buyers were getting 3.1 free- except you!


Cough Up the Money Regards,

[Tony Thomas]

I havent received mine yet, hopefully its worth the wait..

Tony

Alex told me all 3.0 buyers were getting 3.1 free- except you!


Cough Up the Money Regards,

Alex is probably jealous that I offered you the goat, and not him..

[geots]

I havent received mine yet, hopefully its worth the wait..

Tony

Alex told me all 3.0 buyers were getting 3.1 free- except you!


Cough Up the Money Regards,

Alex is probably jealous that I offered you the goat, and not him..

[playwaycool]

Sounds good. It appears to me that Naum 3.0 Generally doing really well in terms of ranking and is currently takes 2nd place in the rating lists that I have viewed recently.


I have a request/desire or wish perhaps someone can fulfill that for me and for other members on this forum to watch. OK it appears to me that some members enjoy testing some programs and post results in Computer Chess Club: Tournaments and Matches section. I have been a member for several years and mostly I'm observer. The tournament results that I have enjoyed mostly are from the following member/moderator Thorsten, he certainly posted some interesting result with good amount of programs in the past!, saddat tournaments is one another member too
Also other tournament testing from Leo at http://wbec-ridderkerk.nl/html/premier.htm and Daniel Charles amazing results and participation test tournament at http://www.superchessengine.com/ Thats about it, others have interesting tournament results and Matches between Rybka vs Naum, Rybka vs Zappa Mexico etc etc... but for me Tournament result is highly preferable too!



However the request/desire or wish (ha!) consist of someone providing us result with a lot of participation much closer to Leo testing tournament but with latest programs version and 64 bit Windows for top programs to get better results than 32 bits speed for top programs! The problem with Leo and Charles tournament is this. Leo have large amount of participation, first 18 are very good programs to see in one tournament, however he is missing Fritz 11 (perhaps because he is using arena) and Hiarcs 12 and other hiarcs version (padeborn?) although he is using hiarcs X50 but it would have been best to test with the latest versions and he is also using old version of Shredder. Obviously updated Hiarcs and Shredder will make huge impact to his tournament! for Daniel Charles test tournaments his 8 cores comp is good to see how well top programs perform along with the 64 bit advantage programs. However the only slight disadvantage for me in my view about his testing is he is missing few good engines like Ktulu and Spike chess program (a program that some of us are observing it's progress in leo test tournament!) (by the way testing can go for test blitz tournament, which might be a good start to share results quite regularly with the members here, to all the way standard time control test tournament!) and perhaps with a good general opening book, I heard perfect 13 opening book is good!

So that being said, Isn't worth testing different parameters for the top programs, perhaps the new Naum 3.1 , Zappa, Hiarcs, and Shredder with the idea in mind to come up with better settings than the default for all 4 or more programs to see how well programs will place and whether a program can edge out Rybka in a big
test tournament with good participation. Here's some proof to what I have seen and what I mean can make an impact for test tournament with good amount of participation in. Dieter Eberle Came up with better settings for Toga than the default and it was a progress against variety of programs opponent, Zied a member on here who actually had his own website and came up with a variety of settings that were better than default settings for Shredder and he also found good setting for Fritz 10, Recently someone came up with better settings for the new hiarcs it seems : http://64.68.157.89/forum/viewtopic.php ... 76&t=20708 I'm certain someone can find better settings for Naum 3.1 as well And testing tournament with good comp such as Quads Intel or AMD sounds good or 8 cores comp, obviously purpose to find how these big program perform



My reaction for such test tournament is a serious test for Rybka and it can actually perform well under the conditions I listed above! It may perform better than any program to receive 1st place still but it will be very close fight, my prediction is that naum or zappa can or even hiarcs can perform better in an open wide/long test tournament assuming they have better settings than default (tested before hand). Rybka in my own opinion remains number 1 program and if we need to make a similar comparison to a human player we can say it might be similar to Vladamir Kramnik where Kramnik is known that he is best suited in a match 1 on 1 and we all witness his winning against Kasparov and Topalov in the world championship, his solid style fit for such matches, however he may not be good tournament player that are able to win a lot of international tournaments from GMs (generally). GM Veselin Topalov for instance is good tournament player because he is risk taker and may end up losing few games and win a whole load of games in arrow to win a tournament something that Kramnik doesn't normally do. Same thing could be said for Rybka, Rybka wins major events but with most or all top programs in a tournament it might not be easy, One program may be perform better because score average among all tournament games. We also witness that Rybka lost major international events to Hiarcs as of recent! What do you guys think?

[Tony Thomas]

However the request/desire or wish (ha!) consist of someone providing us result with a lot of participation much closer to Leo testing tournament but with latest programs version and 64 bit Windows for top programs to get better results than 32 bits speed for top programs! The problem with Leo and Charles tournament is this. Leo have large amount of participation, first 18 are very good programs to see in one tournament, however he is missing Fritz 11 (perhaps because he is using arena) and Hiarcs 12 and other hiarcs version (padeborn?) although he is using hiarcs X50 but it would have been best to test with the latest versions and he is also using old version of Shredder. Obviously updated Hiarcs and Shredder will make huge impact to his tournament!you guys think?

Leo does not use Arena, WBEC stands for winboard engine competition...He is also going through some rough times, hence no new commercial engines. He probably got the longest running tournament on the web, and its the only site I visit daily. In my view its the responsibility of an engine author to make sure that the latest version is running on WBEC...


I received my Naum at 10:30 by the way, I guess Alex is not jealos after all.