Sunday, October 31, 2010

Using Intel© Math Kernel Library in Python

As a way to interact with MKL library from Python is build a shared library from MKL and interface to it using ctypes.

Below is description of the process of creating shared library from MKL.

  1. Download MKL Library from: http://software.intel.com/en-us/articles/non-commercial-software-download/ (for noncommercial use).

  2. Unpack archive and install

  3. Follow this useful article: http://software.intel.com/en-us/articles/using-intel-mkl-in-your-python-programs/


Steps for me:

  1. /tools/environment/mklvarsem64t.sh

  2. cd /tools/builder

  3. $ cat > ring_list
    dsyevr_
    ^D

  4. $ make em64t name=mkl4py export=ring_list

  5. Done. Use mklpy.so and /lib/em64t/libiomp5.so.


MKL online documentation: http://software.intel.com/en-us/articles/intel-math-kernel-library-documentation/. See "Building Custom Shared Object".

Friday, October 8, 2010

Build OpenSIPS deb package from SVN. Step by step.

If you want to build OpenSIPS package for debian/ubuntu from svn, that this is post for you.

First, checkout OpenSIPS sourses from svn as described here.

For example, to download latest 1.6 branch do:
$ svn co https://opensips.svn.sourceforge.net/svnroot/opensips/branches/1.6 opensips

Enter opensips directory and move debian-directory on top-level:
$ cd opensips
$ ln -s packaging/debian debian

Increase the Debian release number:
$ debchange -i

Finally, build package:
$ debuild -us -uc

Debian packaging utilites can be installed with "sudo aptitude install devscripts build-essential fakeroot".