“3 day MATLAB computation done in 4 minutes with a cost of $10” is the message of a recent tweet by the Tampere based Finnish Techila Technologies Ltd folks which first got me interested in their distributed computing solution. The Techila system is a distributed computing middleware and management solution for computing servers, clusters and cloud services. I am mostly interested in the latter variant, more precisely, I decided to have a closer look at Techila in combination with MATLAB on the Google cloud platform which provides a full pay-as-you-go cloud solution together with Google’s state-of-the-art monitoring, logging and diagnostics tools. Sounds exciting? In my eyes it is even more appealing with regard to the fact that Google is offering Google cloud platform education grants which give students free credits and the ability to learn on the platform during their University courses thus having the potential to bring on-demand supercomputing to every student’s desk. Over the weekend, I have tested this framework by studying a commonly used benchmark problem from mathematical finance, namely pricing and sensitivity calculation of a swaption portfolio within a LIBOR market model using Monte Carlo simulation. For students with no or restricted access to the local super computer it is no exaggeration to say that Techila might provide the biggest bang for the buck.
Martin Simon's research blog 