Summer of HPC begins

Summer of HPC begins
Summer of HPC 2020

Well, hello and welcome to the first entry of this year’s Summer of HPC blog. My name is Petar and I would like to introduce myself and share some of my first impressions on SoHPC programme provided by PRACE.

Introduction

As I mentioned before, I am Petar Đekanović and I am born and raised in Belgrade, the capital of a small country located in Southeast Europe called Serbia. I am currently pursuing MSc degree in Software Engineering at University of Belgrade in Serbia.

Profile photo
SoHPC 2020 T-shirt and myself

I found a interest in high performance computing last year as I had attended a course on this topic. I remember being really moved by this new idea of executing the program not sequentially nor concurrently, but fully parallel using multiple computing nodes at the same time and as a consequence getting enormous performance gains.

Although I had an opportunity to use a “super” computer at my school, it was far from equipped to answer the needs of high-demand computational programs. Luckily, at the end of the course my professor shared with the students one HPC programme in which he had participated and, as you may guessed it already, I will be a part of.

And that is how it all started, but enough of me, let us fast-forward to the beginning of the summer of 2020.

Training week

Finally, the summer has started for all participants. But unfortunately, this year is the special one. Due to COVID-19 world pandemic, programme had to shift from its usual in-person format to remote participation.

Despite being online, first day went very well. All of us had an opportunity to meet with each other and exchange our opinions on this new kind of collaboration. We used different online tool such as Mural and Mentimeter to communicate and to learn, which I had not had a chance to use before.

Group photo
Training week’s group “photo” of all attendees and coordinators.

After the first introductory day, we began to explore essential technologies for program parallelization:

  • MPI – high performance Message Passing Interface library
  • OpenMP – shared memory based API for parallelism in C, C++ and Fortran programs
  • CUDA – parallel computing platform for computing on graphical processing units (GPUs)

Using these tools would not be of any help if attendees did not have an access to one of the most powerful clusters in Europe, VSC-3, courtesy of Vienna Scientific Cluster (VSC). Using this cluster gave all of us a glimpse to what real state-of-the-art computer can do and with extensive help of all staff and coordinators, we successfully gained necessary training we needed.

Also, one of advantages of online training was that this year, for the first time, all applicants could join the training and benefit from it same as participants, which had been really amazing.

Although I did not get an experience that I would being there in person, the training week gave me a different view on how things could be done and definitely encouraged and as well prepared me for the next challenge – team project with my fellow colleague Roberto Rocco (if you are interested, you can take a look on his side of the story on his blog).

Conclusion

As you can probably suspect, this has only been a first look on how internship is being organized and my personal view on it so far. Thank you for stopping by and do not worry – new posts will come soon! Stay tuned!

Please follow and like us:
error
Tagged with: , , , ,

Leave a Reply

Your email address will not be published. Required fields are marked *

*

This site uses Akismet to reduce spam. Learn how your comment data is processed.