Playing Well with Others in HPC


Playing Well with Others in HPC

Einstein spent seven years in a patent office as an evaluator, an environment he called his “worldly cloister where I hatched my most beautiful ideas.” Dividing his day between sleep, scientific exploration, and office work, he spent his office hours evaluating novel ideas and his off hours contemplating the mysteries of the universe. I imagine this balanced use of Einstein’s brain power most likely resulted in insights and discovery that would not have occurred if he had locked himself away with a pile of physics books for 16 hours a day.

We often imagine scientific discovery taking place in isolated laboratories, the years’ long culmination of a single, dedicated researcher. The truth that I have found is it results from varied points of view, collaboration, and the insights gained from diverse backgrounds. As a Ph.D. working for a corporation, which in turn supports the DoD, I feel like I sit where industry, government, and academia meet: the perfect place for collaboration and scientific insight.

So how do we foster that magic mix? How do we intentionally create opportunities for individuals from varied organizations and backgrounds to discuss problems in extreme scale computing?

The Department of Energy (DOE) has been considering this since the 1970s when they created the Computer Graphics Forum (CGF). CGF spans the wider array of analysis, visualization, analytics, and even machine learning – anything that improves our understanding of data. Don't let the name fool you; "Computer Graphics Forum" is anachronistic, dating back to the 70s, but no one wants to change it. Nor is it restricted to DOE – it hasn’t been for decades. Government-affiliated attendees are invited, so it facilitates better collaboration between primarily DOE, the National Science Foundation, and DoD — those that do extreme-scale computing for U.S. government priorities.

Playing Well With Others in HPC Blog Post Lawrence Livermore collaboration event: Source DOE Photos by Randy Wong/LLNL

Several of my most meaningful projects came from ideas discussed at that meeting. The DOE often uses it to evaluate analysis/viz technical priorities to fund in coming years. As such, each of the agencies that attend can identify complementary topics that will benefit multiple agencies while still being tailored to their own needs. Each can also share their most bleeding-edge developments for the benefit of the others.

The DOE is not the only organization creating “cloisters” of collaboration. A large part of my job is enabling researchers at national laboratories and connecting them to scientists from academia. Join that with the insights that my fellow industry-based computational scientists and I bring to the table, and we have a really powerful convergence of ideas and talent.

We scientists sometimes grumble when pulled from our terminals, but I’ve learned from Einstein and personal experience, that a little time in the world can pay dividends in achieving meaningful breakthroughs.

Share this Post:

Posted by Sean Ziegeler

I coordinate computational scientists for the DoD HPCMP PETTT program in the Advanced Computational Environments area. My areas of interest include parallel I/O, performance optimization, performance profiling, visualization and analysis, emerging computational technology, and general parallel computing. I have been involved with the DoD HPCMP in various capacities for over 20 years. I hold a Ph.D. in Computer Engineering from Mississippi State University.