Numexpr is a fast numerical expression evaluator for NumPy. With it,
expressions that operate on arrays (like "3*a+4*b") are accelerated
and use less memory than doing the same calculation in Python.
It wears multi-threaded capabilities, as well as support for Intel's
VML library, which allows for squeezing the last drop of performance
out of your multi-core processors.
What's new
==========
In this release, better docstrings for `evaluate` and reduction
methods (`sum`, `prod`) is in place. Also, compatibility with Python
2.5 has been restored (2.4 is definitely not supported anymore).
In case you want to know more in detail what has changed in this
version, see:
http://code.google.com/p/numexpr/wiki/ReleaseNotes
or have a look at RELEASE_NOTES.txt in the tarball.
Where I can find Numexpr?
=========================
The project is hosted at Google code in:
http://code.google.com/p/numexpr/
You can get the packages from PyPI as well:
http://pypi.python.org/pypi/numexpr
Share your experience
=====================
Let us know of any bugs, suggestions, gripes, kudos, etc. you may
have.
Enjoy!
--
Francesc Alted
On Sun, Jan 8, 2012 at 11:49, Francesc Alted <fal...@gmail.com> wrote:
> What's new
> ==========
>
> In this release, better docstrings for `evaluate` and reduction
> methods (`sum`, `prod`) is in place.
I always wondered why they weren't documented... IIRC, there are a few
more undocumented functions: ones_like, copy and fmod.
It would be nice if you could build windows packages for this release.
I can build 32-bit packages (with MingW) if that helps, but can't do
64-bit ones without much hassle (if at all).
Thanks,
--
Gaëtan de Menten
Well, it is not that I have all the time in the world, but anyway :)
> On Sun, Jan 8, 2012 at 11:49, Francesc Alted <fal...@gmail.com> wrote:
>
>> What's new
>> ==========
>>
>> In this release, better docstrings for `evaluate` and reduction
>> methods (`sum`, `prod`) is in place.
>
> I always wondered why they weren't documented... IIRC, there are a few
> more undocumented functions: ones_like, copy and fmod.
You are right. Please feel free to document these properly.
> It would be nice if you could build windows packages for this release.
> I can build 32-bit packages (with MingW) if that helps, but can't do
> 64-bit ones without much hassle (if at all).
Hmm, I think people already knows this:
http://www.lfd.uci.edu/~gohlke/pythonlibs
which has the advantage that Christof is linking against MKL.
--
Francesc Alted
I did know about those, but I don't think we can assume everybody does.
> which has the advantage that Christof is linking against MKL.
which is precisely the reason I need other packages because I need to
be able to distribute those packages and I don't have the rights to
distribute anything linked to the MKL.
--
Gaëtan de Menten