MathGroup Archive 2013

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Limiting the number of cores used by a single kernel?

  • To: mathgroup at smc.vnet.net
  • Subject: [mg130639] Re: Limiting the number of cores used by a single kernel?
  • From: Bill Rowe <readnews at sbcglobal.net>
  • Date: Sun, 28 Apr 2013 00:59:59 -0400 (EDT)
  • Delivered-to: l-mathgroup@mail-archive0.wolfram.com
  • Delivered-to: l-mathgroup@wolfram.com
  • Delivered-to: mathgroup-outx@smc.vnet.net
  • Delivered-to: mathgroup-newsendx@smc.vnet.net

On 4/26/13 at 11:10 PM, georgevw3 at mac.com (George Woodrow III) wrote:

>On Apr 26, 2013, at 4:25 AM, bradc355113 at yahoo.com wrote:

>>I've noticed recently (beginning in v8 or so) that some functions seem
>>to make use of multiple threads internally. For example, Resolve and
>>some linear algebra functions will push my CPU usage up to 400% while
>>they run, even with just one kernel.

>>Is there a way to limit this?

>In Mathematica 9 (and earlier for all I know), in Preferences, there
>is a panel labeled Parallel.

>There is a section for local kernels, and you can manually set the
>number of kernels.

>Setting this to 1 should solve your problem.

I don't think this will solve the original poster's problem.
This would prevent multiple kernels from being launched and
consequently prevent Mathematica's ability to do parallel
processing. But it would not prevent multiple threads from being
launched which is something different than parallel processing.
Note the original poster stated the problem occurred when only
one kernel was launched.

Frankly, I don't know how to keep Mathematica from launching
multiple threads or even what might cause Mathematica to do so.
That quite likely has to be restricted by some system setting
independent of Mathematica.





  • Prev by Date: Re: Mathematica-assisted learning was .. Re: Speak errors
  • Next by Date: how to generate piecewise function from two lists
  • Previous by thread: Re: Limiting the number of cores used by a single kernel?
  • Next by thread: Convolve: Different Looking Results. (2)