Re: [OctDev] random numbers for distributed computations

[Paul Kienzle]

On Apr 11, 2005, at 9:17 AM, Michael Creel wrote:

> One last note, the reference
>   
> http://www.cosc.canterbury.ac.nz/research/reports/TechReps/2005/ 
> tr_0503.pdf
> leads me to believe that the method used by the octave-forge rand.cc is
> reliable. Thanks to everyone for the information, Michael

Thanks for the link.

They report a 4s startup time for MT on a 2.4 GHz machine.  I can
assure you this is not the case for randn on octave-forge.  We
are definitely not doing what the code in Dynamic Creation
Mersenne Twister (DC) is doing.

Given the long startup time for DC, you need 70 M random numbers
just to break even with the next fastest.  Depending on what you
are doing, you may be better off using SPRNG's multiplicative
lagged fibonacci generator (MLFG).

- Paul



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------