By Strahinja Trecakov, Systems Developer, Information and Communication Technologies at New Mexico State University (NMSU)
Well, I finally made it to SC!
For almost three years, I have wanted to attend the worlds biggest high-performance computing (HPC) conference, but it hasn’t been possible, until 2019.
In July, a colleague recommended the STEM-Trek organization and its programs that help HPC and data specialists attend high-tech conferences. I applied, and few days later, I was selected.
My travel grant included a full tech program registration (workshops and tutorials), and accommodation. My out-of-pocket was minimal, and flights from NM to DEN are usually reasonable. I lodged in STEM-Trek’s “Ubuntu House@SC” — a beautiful, new 3BR/3BA AirBnB condo where three stayed one week for what it would have cost for a single room in a downtown hotel. Two of us shared an Uber or Lyft each day, and with DoorDash, who needs room service?
HPC systems administrators at academic institutions face many roadblocks, and travel is a luxury few can afford these days. Many of our challenges stem from having to educate ourselves about different aspects of supporting hardware and software needed by research scientists who rely on us to help. But the technology is always changing, and it’s difficult to bring ourselves up-to-speed. Conferences are an important source for much-needed training. I not only learned a lot at SC19, I learned from people I met there, and will probably continue to learn from some of them. Professional networking at SC is as important as the technical training you’ll find there.
As for highlights, attending tutorials on the Spack package manager and GPU programming helped me learn more about these two important subjects. At NMSU, we use Spack to manage our software packages, so those lessons were quick to apply on the job. The session on GPU programming taught me a few new tricks; I now feel more confident helping our researchers leverage accelerators to their full potential. Additional sessions that I found useful, and could immediately apply upon returning, were Open On Demand (OOD), and NSF-XSEDE’s Metrics On Demand (XDMoD).
The rest of the conference was really busy. I attended many Birds-of-a-Feather (BoF) sessions, and learned a lot about different HPC tools. Moreover, I listened to talks based on research papers and novel approaches. The expo was mind-blowing since there were over 300 exhibitors; if you wanted to learn about them and their new technologies, you could spend most of your time there–in fact, some from our group only had exhibitor passes, and they found that experience, alone, to be beneficial. On the other hand, every day after the conference doors close, there were many after-hours events. If you want the full SC experience, it’s advisable to to budget your time and energy so that you can attend some of those, too.
To sum up my week at SC19, I would like to break it down into stats: Six days, 13,950 attendees, 300+ exhibitors, 500+ volunteers, etc. It’s the biggest conference I have ever attended, and is a great place to learn, share and connect. Interesting fact: if you only meet one percent of SC19 attendees, it is still 139 new people. Amazing, right?
With donations from SC19, HPCNow!, Google and others, I was one of 16 who could attend this amazing conference as a STEM-Trek participant. Many thanks to all who made it possible!