Cyclone 1.0 beta

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
mclane
Posts: 18956
Joined: Thu Mar 09, 2006 6:40 pm
Location: US of Europe, germany
Full name: Thorsten Czub

Re: Cyclone 1.0 beta

Post by mclane »

as far as i know you can enter the NPS in the UCI-parameters,
tell the engine which time control you use and it will calculate
automatically the size of the Hash.

e.g. lets say it is doing 500.000 NPS.
and you play 40/10 , that is average 15", enter this in the UCI-parameters of cyclone and you will see it uses MORE than the 64 MB hash.
User avatar
Denis P. Mendoza
Posts: 415
Joined: Fri Dec 15, 2006 9:46 pm
Location: Philippines

Re: Cyclone 1.0 beta

Post by Denis P. Mendoza »

And adding to that, here's Norm's other additional notes sent to me that could help in the issue:
One more thing…cyclone includes speedtest
Simply type go speedtest at the console windows to see an nps report.
Don’t know if this will help you or not, I’m sure you have your own manner to determine executable speed.
If you want to test cyclone against another engine, the dynamic hash needs this nps info.

For example:
At the console, type go speedtest
Note the nps measurement
Enter this in the GUI config option ‘nodes per second’
Calculate how many seconds the engine will think per move for the particular time control.

i.e. if the level is 40/40, then it’s 60 seconds per move
enter this the GUI config option ‘seconds per move’

cyclone will then calculate and set the optimal hash size for that computer and that time control
kranium
Posts: 2130
Joined: Thu May 29, 2008 10:43 am

Re: Cyclone 1.0 beta

Post by kranium »

David Dahlem wrote:Hi Norman

Is there a hash limit for Cyclone? In the Arena GUI, in the configuration dialog window, if i set hash to 128 mb or 256 mb, According to Arena and the Windows Task Manager, Cyclone is only using 64 mb.

Thanks
Dave
Hi Dave-

min hash = 8MB
max hash = 1024 MB
the default is 5 seconds per move at 400000 nps = ~ 64MB hash

i'm busy at this very minute creating a rather large document detailing the use of all the different UCI parameters. it will soon be included in the download archive.

in the meantime, there is a page on the website that details dynamic hash and speedtest

here is a link:
http://www.xyclops.com/dynamic_hash.html

Norm
User avatar
Ovyron
Posts: 4562
Joined: Tue Jul 03, 2007 4:30 am

Re: Cyclone 1.0 beta

Post by Ovyron »

David Dahlem wrote:Hi Norman

Is there a hash limit for Cyclone? In the Arena GUI, in the configuration dialog window, if i set hash to 128 mb or 256 mb, According to Arena and the Windows Task Manager, Cyclone is only using 64 mb.

Thanks
Dave
I set it to 16MB and it was using 64MB.
kranium
Posts: 2130
Joined: Thu May 29, 2008 10:43 am

Re: Cyclone 1.0 beta

Post by kranium »

Ovyron wrote:
David Dahlem wrote:Hi Norman

Is there a hash limit for Cyclone? In the Arena GUI, in the configuration dialog window, if i set hash to 128 mb or 256 mb, According to Arena and the Windows Task Manager, Cyclone is only using 64 mb.

Thanks
Dave
I set it to 16MB and it was using 64MB.
Hi Ulysses
please verify that you have dynamic hash turned off via the UCI options configuration menu.
in Arena this can be found via engines->cyclone->configure.

Norm
User avatar
Ovyron
Posts: 4562
Joined: Tue Jul 03, 2007 4:30 am

Re: Cyclone 1.0 beta

Post by Ovyron »

Ok, it was on, I have switched it off. Should I rerun the games that I've played with dynamic hash on, this time switching it off?
kranium
Posts: 2130
Joined: Thu May 29, 2008 10:43 am

Re: Cyclone 1.0 beta

Post by kranium »

if you are seeking test results at 128MB, but the games were played using 64MB, then yes i suppose you need to start again. although it's certainy unclear if the extra memory would hurt or help...it really depends on the time control.

there more info here:
http://www.xyclops.com/dynamic_hash.html and also in the new cyclone UCI options doc, now available for download.

regards-
Norm
Henrik Dinesen
Posts: 877
Joined: Wed Mar 08, 2006 9:52 pm
Location: Denmark

Re: Cyclone 1.0 beta

Post by Henrik Dinesen »

Hi Norman,

I'm wondering what happened to the MPV code of Toga. I do realize that some versions of Toga doesn't have it anymore, but it was implemented at some point in Thomas code.

To those who enjoys analysing, MPV can be a very nice feature.
Henrik
kranium
Posts: 2130
Joined: Thu May 29, 2008 10:43 am

Re: Cyclone 1.0 beta

Post by kranium »

Henrik Dinesen wrote:Hi Norman,

I'm wondering what happened to the MPV code of Toga. I do realize that some versions of Toga doesn't have it anymore, but it was implemented at some point in Thomas code.

To those who enjoys analysing, MPV can be a very nice feature.
Hi Henrik,

MultiPV was introduced in early versions of Toga (or was it GambitFruit? not sure here, the two share a lot of code), but as far as I know it was never fully functional. (please correct me if i'm mistaken)

in the cyclone10_beta.txt included with the distribution, it's listed as #1 on my 'to do list'. Although implementing it is likely not trivial , i do hope to get it functional in future versions.

Norm
Uri Blass
Posts: 11161
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Cyclone 1.0 beta

Post by Uri Blass »

kranium wrote:Hi all-

i have been working non-stop on Cyclone 1.0 beta, and it is now ready to be released. i've was in contact yesterday with Werner Schule from CEGT about the release, and have asked Jim Ablett to produce optimized compiles for distribution. i'm now waiting for an answer from Jim. If there is anyone else that has the time, and is willing to do this, please contact me asap.

i would do it myself, but my compiler skills and resources are limited. As soon as I have these optimized executables, I will release a package as open-source on the www.xyclOps.com website, or perhaps ridderkerk.

as a courtesy, i'm attaching the readme.txt here in an effort to provide some information. Cyclone is based on Fruit/Toga, but uses many ideas, features, and improvements from xyclOps. my initial testing is quite positive and encouraging, but i won't cite any ELO numbers here, i prefer to leave that to you all, and the experts at CEGT, CCRL, etc.

Best Regards-
Norm Schmidt
I wonder why no results of this program until now.
I looked at the CEGT forum and found no subject for Cyclone 1.0 beta
I also found nothing in the ccrl live rating list about Cyclone

Do they wait Jim Ablett to produce optimized compiled version?

Uri