[SciPy-User] multiple cores

[David Goldsmith]

Hi, Carlton.  This is not an area of expertise of mine, but I took the
liberty of Googling the topic and here's the top hit:
http://scipy.github.io/old-wiki/pages/ParallelProgramming.  The top line
is: "This is an archival dump of old wiki content --- see scipy.org for
current material," so I clicked on the link to go to the scipy home page
and repeated my search there...and the top hit was the same page. :-(
 Scrolling down in these search results, I see some more-or-less relevant
links, but nothing general, so I started reading the "old" page.  These are
the highlights, IMO:

"numpy/scipy are not perfect in this area, but there are some things you
can do."

"The best way to make use of a parallel processing system depend on the
task you're doing and on the parallel system you're using."  (I regard this
as particularly pertinent, absent knowledge of your system, as one thing I
do know is that Windows is, or at least was--I stopped using Windoze with
the introduction of 8--particularly "protective" of the allocation of
processors; I'm less familiar with the "friendliness" of Posix flavors,
including Linux and Mac.)

And finally:

""premature optimization is the root of all evil
<http://scipy.github.io/old-wiki/external.html?link=http://www.acm.org/ubiquity/views/v7i24_fallacy.html>"...Get
your code working first, before even thinking about parallelization. Then
ask yourself whether your code actually needs to be any faster. Don't
embark on the bug-strewn path of parallelization unless you have to."

Wiser words are rarely spoken.  Nevertheless, though billed as old, that
document _appears_ to be the most current, general document on parallel
programming w/ scipy, so it appears to be the place to start (and it _does_
have specific suggestions on how and when to parallelize scipy code).

I'd hope that more expert voices will chime in, but _sometimes_ that takes
a little while, so I wanted to give you something to mull in the meantime.

FWIW,

DLG

On Fri, May 5, 2017 at 4:18 PM, Carlton Banks <noflaco at gmail.com> wrote:

> Anyone are of whether SciPy
> <https://docs.scipy.org/doc/scipy-0.15.1/reference/generated/scipy.optimize.differential_evolution.html> use
> multiple cores?
>
> if not is there a possibility of this?
>
> _______________________________________________
> SciPy-User mailing list
> SciPy-User at python.org
> https://mail.python.org/mailman/listinfo/scipy-user
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/scipy-user/attachments/20170505/c01fb8d5/attachment.html>