== Tutors ==

Francesc Alted (francesc at blosc dot org)

== Topics covered ==

  * Memory efficient programming
  * Compression
  * multiprocessing with numexpr
  * pandas (short overview)
  * bcolz (compressed data management)

== Lecture ==

The slides are {{starving_cpu:The_Starving_CPU_Problem_-_Split_2014.pdf|here}}.

== Exercises ==

=== Using your local machine ===

The exercises are {{starving_cpu:starving-cpu-exercises.tar.gz|here}}.  Just download them (they are around 50 MB large, so be patient), unpack and launch ipython notebook:
<code bash>
$ tar xvfz starving-cpu-exercises.tar.gz
$ cd starving-cpu-exercises
$ ipython notebook</code>

And do the exercises by following the guidelines in the "Split14-Starving-CPUs-Exercises.ipynb" notebook.

=== Using the Big Iron machine for heavy-duty exercises ===

Also, the [[http://www.gc3.uzh.ch/|GC3 group at University of Zurich]] generously provided a NUMA 48-core machine (SGI SV 100), and its usage is very much recommended for the multiprocessor and data storage exercises (exercises 5 and on).  For using it, just download the next {{https://python.g-node.org/wiki/_media/starving_cpu/split14-starving-cpus-exercises-bi.ipynb|notebook}} and copy it to the Big Iron machine with:
<code bash>
$ scp split14-starving-cpus-exercises-bi.ipynb USER@login.gc3.uzh.ch:</code>
(replace USER with your actual username).

Now, please launch an IPython kernel in the remote machine by following the next {{https://python.g-node.org/wiki/setup|procedure}}, and you are ready to go.

== Solution for the exercises ==

You can find them {{https://python.g-node.org/wiki/_media/starving_cpu/split14-starving-cpus-solutions.ipynb|here}}.