<div dir="ltr"><div class="gmail_default" style="font-family:arial,helvetica,sans-serif">







<p class="gmail-p1"><span class="gmail-s1">=========================</span></p>
<p class="gmail-p1"><span class="gmail-s1"> Announcing Numexpr 2.6.2</span></p>
<p class="gmail-p1"><span class="gmail-s1">=========================</span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1">What's new</span></p>
<p class="gmail-p1"><span class="gmail-s1">==========</span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1">This is a maintenance release that fixes several issues, with special</span></p>
<p class="gmail-p1"><span class="gmail-s1">emphasis in keeping compatibility with newer NumPy versions.  Also,</span></p>
<p class="gmail-p1"><span class="gmail-s1">initial support for POWER processors is here.  Thanks to Oleksandr</span></p>
<p class="gmail-p1"><span class="gmail-s1">Pavlyk, Alexander Shadchin, Breno Leitao, Fernando Seiti Furusato and</span></p>
<p class="gmail-p1"><span class="gmail-s1">Antonio Valentino for their nice contributions.</span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1">In case you want to know more in detail what has changed in this</span></p>
<p class="gmail-p1"><span class="gmail-s1">version, see:</span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1"><a href="https://github.com/pydata/numexpr/blob/master/RELEASE_NOTES.rst">https://github.com/pydata/numexpr/blob/master/RELEASE_NOTES.rst</a></span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1">What's Numexpr</span></p>
<p class="gmail-p1"><span class="gmail-s1">==============</span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1">Numexpr is a fast numerical expression evaluator for NumPy.  With it,</span></p>
<p class="gmail-p1"><span class="gmail-s1">expressions that operate on arrays (like "3*a+4*b") are accelerated</span></p>
<p class="gmail-p1"><span class="gmail-s1">and use less memory than doing the same calculation in Python.</span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1">It wears multi-threaded capabilities, as well as support for Intel's</span></p>
<p class="gmail-p1"><span class="gmail-s1">MKL (Math Kernel Library), which allows an extremely fast evaluation</span></p>
<p class="gmail-p1"><span class="gmail-s1">of transcendental functions (sin, cos, tan, exp, log...) while</span></p>
<p class="gmail-p1"><span class="gmail-s1">squeezing the last drop of performance out of your multi-core</span></p>
<p class="gmail-p1"><span class="gmail-s1">processors.  Look here for a some benchmarks of numexpr using MKL:</span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1"><a href="https://github.com/pydata/numexpr/wiki/NumexprMKL">https://github.com/pydata/numexpr/wiki/NumexprMKL</a></span></p>
<p class="gmail-p2"><span class="gmail-s1"></span><br></p>
<p class="gmail-p1"><span class="gmail-s1">Its only dependency is NumPy (MKL is optional), so it works well as an</span></p>
<p class="gmail-p1"><span class="gmail-s1">easy-to-deploy, easy-to-use, computational engine for projects that</span></p>
<p class="gmail-p1"><span class="gmail-s1">don't want to adopt other solutions requiring more heavy dependencies.</span></p><p class="gmail-p1"><span class="gmail-s1"><br></span></p><p class="gmail-p1"><span class="gmail-s1">Where I can find Numexpr?</span></p><p class="gmail-p1"><span class="gmail-s1">=========================</span></p><p class="gmail-p2"><span class="gmail-s1"></span><br></p><p class="gmail-p1"><span class="gmail-s1">The project is hosted at GitHub in:</span></p><p class="gmail-p2"><span class="gmail-s1"></span><br></p><p class="gmail-p1"><span class="gmail-s1"><a href="https://github.com/pydata/numexpr">https://github.com/pydata/numexpr</a></span></p><p class="gmail-p2"><span class="gmail-s1"></span><br></p><p class="gmail-p1"><span class="gmail-s1">You can get the packages from PyPI as well (but not for RC releases):</span></p><p class="gmail-p2"><span class="gmail-s1"></span><br></p><p class="gmail-p1"><span class="gmail-s1"><a href="http://pypi.python.org/pypi/numexpr">http://pypi.python.org/pypi/numexpr</a></span></p><p class="gmail-p2"><span class="gmail-s1"></span><br></p><p class="gmail-p1"><span class="gmail-s1">Share your experience</span></p><p class="gmail-p1"><span class="gmail-s1">=====================</span></p><p class="gmail-p2"><span class="gmail-s1"></span><br></p><p class="gmail-p1"><span class="gmail-s1">Let us know of any bugs, suggestions, gripes, kudos, etc. you may</span></p><p class="gmail-p1"><span class="gmail-s1">have.</span></p><p class="gmail-p2"><br><span class="gmail-s1"></span></p><p class="gmail-p1">

























</p><p class="gmail-p1"><span class="gmail-s1">Enjoy data!</span></p></div><div><br></div>-- <br><div class="gmail_signature">Francesc Alted</div>
</div>