[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Could eig() be outperformed by MKL numpy?
|
From: |
Klonuo Umom |
|
Subject: |
Re: Could eig() be outperformed by MKL numpy? |
|
Date: |
Wed, 18 May 2011 00:29:11 +0200 |
> 1. Simple dot product - dot(a,b)
>
> ======================
> Octave : 0.031
> Cygwin/ATLAS : 0.843
> Windows/ATLAS* : 1.29
> Windows/MKL : 2.37
> ----------------------
> *from official numpy binary installer
As shown by member on SciPy ML this is wrong table, because dot() in Octave
and NumPy are different operations. In Octave, result is vector. Right operator
in Octave would be '*', and this is updated table:
======================
Octave : 0.766
Cygwin/ATLAS : 0.843
Windows/ATLAS* : 1.29
Windows/MKL : 2.37
----------------------
Now this are expected results, and nothing special to be looked at.
I'm posting it only as correction
Sorry for previous misinformation
However, question about overperfomed eig() is still valid