Re: st: Stata/MP really exploiting my processor's 4 cores?

[Austin Nichols <austinnichols@gmail.com>]

Patrik Morgetz <pmorgetz@gmail.com>:
See for ideas e.g.
http://www.stata.com/statalist/archive/2009-07/msg01281.html
http://www.stata.com/statalist/archive/2011-07/msg00086.html
http://www.stata.com/statalist/archive/2011-03/msg01018.html

On Mon, Sep 12, 2011 at 4:18 AM, Patrik Morgetz <pmorgetz@gmail.com> wrote:
> Hello everyone
>
> In my university, some time ago, the IT people changed the computers,
> and I got a laptop with a pretty good processor -Intel® Core™
> i7-2630QM Processor (6M Cache, 2.00 GHz, 4 cores, 8 threads)-, 8GB ram
> memory and running windows 7 (64-bits version).
>
> These days I am running quite intensive processes in Stata
> (bootstrapping and multinomial probit, both on large datasets), and it
> takes a lot of time to run. So, since I have a capable processor to
> run it, I asked the IT people to install Stata/MP, hoping that it will
> run so much faster because the Stata/MP performance report
> (http://www.stata.com/statamp/statamp.pdf) explains that "it
> distributes many of Stata’s most computationally demanding tasks
> across all the cores in your computer and thereby runs faster—much
> faster".
>
> At first glance, it seems that Stata/MP was correctly installed (it
> works, and at start up it shows the "MP - Parallel Edition" label),
> however, I did not notice a dramatic improvement in the performance,
> so I am wondering whether it is really taking advantage of the
> parallel processing capabilities of my processor. To make sure about
> it, I closed every other program and put Stata to run the time
> consuming bootstrapping process, then I opened the Windows Task
> Manager to take a look to the performance tab where it shows the CPU
> usage (it shows 8 slots, I guess, 2 threads per every core), and I
> expected to see a high CPU usage in every slot there, because it would
> suggest it is actually parallel processing. But it didn't !!!, it just
> shows one of the CPU-8 slots at its maximum usage, and the others at
> very low usage levels. I then tried running other Stata commands that
> seems to be processing intensive (margins, roctab, roccomp, asmprobit,
> mprobit), and the result was the same (one CPU-slot at top usage and
> the others at very low levels).
>
> So, I am afraid Stata/MP is not really taking advantage of the
> parallel processing capabilities of my processor. Am I right?, is it
> correct to determine parallel processing the way I am doing it?
> And most importantly, if I am right and Stata/MP is not parallel
> processing, how may I fix it? (I guess it is an installation problem,
> but no idea how to deal with it). I am using Stata/MP version 11.2
>
> So thank you very much for your help with this issue.
>
> Regards,
>
> Patrik Morgetz

*
*   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/