Last month was a busy one. On February 11th I started working on a project called Galileu. This project is being implemented by a joint venture between Petrobras and a group of universities. Amongst those universities is the Aeronautics Technological Institute (ITA).

I got such job opportunity through ITA's mechanical engineering department, more specifically the Computational Transport Phenomena Laboratory. I received an email from its group leader Marcelo de Lemos. I answered the email right way sending him my resume. They needed someone with expertise in programming languages and high performance computing (HPC). I was eagerly waiting a job opportunity because I finished the computer engineering course and there was no place to work. This opportunity was just what I needed.

My work is really exciting. Firstly I was given material about the message passing interface (MPI) and the Rocks clusters distribution for the Linux platform. After that I started playing with the Linux cluster, which was already installed. Who installed it was a great friend I met at the lab. His name is Arkady Petchenko.

As a lab we experiment with the available HPC platforms. The last two weeks I've been playing with the Windows Computer Cluster Server 2003. I was responsible for installing the Windows cluster.

I'm also doing development of code in C/C++ and Fortran. I compile and build applications that can explore the full potential of a cluster through the MPI API.

It's fantastic to see how a parallel program performs on different numbers of processors.

In future posts I'd like to discuss about the aspects of the technologies I work with. It's great stuff! I love programming languages and the parallel computing world is a fascinating one. With the increase of the number of processors in a single chip, we're going to see a vast amount of parallelized code being executed on distributed computing systems. Parallel APIs as the Microsoft Parallel Extensions for .NET framework are evolving rapidly. This will for sure change the way we think when working with code and specially with clusters and supercomputers.