[Numpy-discussion] numpy-1.6.1.win32-py3.2.exe (md5) won't install

[Charles R Harris]

On Tue, Aug 30, 2011 at 10:27 AM, Richard D. Moores <rdmoores at gmail.com>wrote:

> On Tue, Aug 30, 2011 at 09:09, Charles R Harris
> <charlesr.harris at gmail.com> wrote:
> >
> >
> > On Tue, Aug 30, 2011 at 10:01 AM, Richard D. Moores <rdmoores at gmail.com>
> > wrote:
> >>
> >> On Tue, Aug 30, 2011 at 08:51, Olivier Delalleau <shish at keba.be> wrote:
> >> >
> >> > 2011/8/30 Richard D. Moores <rdmoores at gmail.com>
> >> >>
> >> >> Is it possible to install 32-bit
> >> >> Python 3.2 on 64-bit Win 7 (you seem to have done so), so I could use
> >> >> numpy?
> >> >>
> >> >
> >> > Yes you can insteall Python 32 bit on 64 bit Windows.
> >>
> >> Thanks. Would doing so leave my 64-bit Python 3.2 intact, so I could
> >> switch to the 32-bit only to install and use numpy?
> >>
> >
> > You might want to try the win64 packages here.
> >
> > Chuck
>
> Thanks Chuck! I downloaded
> numpy-unoptimized-1.6.1.win-amd64-py3.2.exe.  numpy is now installed
> for 64-bit Python 3.21
>
> But what are the implications of "unoptimized"?
>
>
Array operations will be slower. The optimized versions will be faster
because they are linked to the highly optimized and tuned Intel MKL library
rather than the fallback code included in numpy. If you have a lot of big
arrays the speed difference will be significant. For small arrays call
overhead tends to dominate and there isn't that much difference.

You might want to download ipython and matplotlib also so that you have the
basic numpy stack.

Chuck
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/numpy-discussion/attachments/20110830/a86b23df/attachment.html>