Re: st: RE: Will Stata/MP speed up running multiple dofiles inbatch mode?

[Neil Shephard <nshephard@gmail.com>]

On Thu, Jan 20, 2011 at 10:25 AM, Alex Eapen <alex.eapen@sydney.edu.au> wrote:
> Thanks Neil. That clarifies it for me. The commands I intend to run in each dofile are -summ- and  -regress-. So by your reasoning  I will see improvements when I upgrade to Stata/MP.

Hmm, thats not really what I was trying to convey.  I don't have a
version of Stata/MP to test this with, but in my opinion because you
will be starting multiple instances of Stata/MP (one to run each batch
file) you are unlikely to see any great improvement in the speed at
which they all complete because how can one instance know that another
instance is already running and take advantage of the hyperthreading
that is used to spread jobs across the multiple cores of your CPU?
Its one instance of Stata/MP that takes advantage of the multiple
cores in your CPU because its be re-written to use hyper-threading.

You'll see an improvement if, and only if, running bigjob1.do runs
faster under Stata/MP than it does under Stata/SE because _each_ batch
job will complete faster (see the official Stata report I linked for
details and quantification of this).

Perhaps some background on multi-core processors and hyper-threading
would be informative
http://en.wikipedia.org/wiki/Multi-core_%28computing%29 and
http://en.wikipedia.org/wiki/Hyper-threading would make good.

Neil

-- 
“Truth in science can be defined as the working hypothesis best suited
to open the way to the next better one.” - Konrad Lorenz

Email - nshephard@gmail.com
Website - http://kimura.no-ip.org/
Photos - http://www.flickr.com/photos/slackline/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/