Senior Scientist, GE Global Research
While building supercomputing systems is awesome in itself, what’s done using those systems is often where the real “wow factor” is. No one knows this better than Masako Yamada, a Physicist, Sr. Scientist and lead principal investigator for GE Global Research, who was the leader of the GE Global Research group that made the Oak Ridge National Lab’s (ORNL) Cray “Titan” sing last year. What is she and her group up to this year? We caught up with Yamada to ask her.
HPCwire: During your career with GE, you’ve led a number of research projects involving high performance computing across a broad spectrum, including a project involving the Cray Titan. Can you tell us a little about the work you were doing with the Titan, and what you were able to get accomplished?
Masako Yamada: We have been using Titan at Oak Ridge National Lab to model the behavior of water droplets freezing on a variety of surfaces. Our ultimate goal is to guide the design of new anti-icing surfaces, which could potentially enable wind turbines to run more efficiently in cold climate regions The HPC work complements the experimental work that GE Global Research has been pursing for several years.
The simulation team has already achieved our initial goal of replicating lab observations. Our particular value-add has been that we have been able to monitor the re-arrangement of each water molecule as the droplet transitions from liquid to ice. By understanding the fundamental mechanisms that give rise to ice formation, we can work toward ways to avoid it.
HPCwire: Molecular dynamics have been a large part of your life, and you’ve seen the field grow in tremendous leaps and bounds as the computing power has increased. What sort of things do you think further HPC innovation will bring to your abilities? What kinds of problems are you hoping to solve?
Masako Yamada: Incredible breakthroughs in the molecular dynamics field have enabled billion-atom simulations or millisecond simulations to become a reality… but usually not at the same time. Even these cutting-edge molecular dynamics simulations are simple, tiny and short compared to many real-life challenges that impact our daily lives. I’m hoping HPC can someday be used to address corrosion or disease progress, where many moles of molecules (mole = 6 x 10^23) need to be monitored over seconds, hours or even years. We’re at least 15 orders of magnitude away, so there’s a lot of opportunity to innovate hardware, acceleration algorithms, data analysis, data transfer and storage. Knowing which tools to use for a given problem is one of the subtleties and skills of being a good computational scientist.
HPCwire: You’ve mentioned to us that you think an industry shift towards a more holistic view of the impact of HPC is important, including the value of adopting hybrid accelerators. Why do you see this as such an important trend? Are there any other areas you see as particularly important?
Masako Yamada: It’s human nature to use metrics such as ROI to quantify progress. But a holistic approach becomes necessary when multiple variables are intertwined, when both costs and benefits may be hidden, and when results may take years to materialize. For example, my collaborator Mike Brown (Oak Ridge National Lab) and I spent about a week porting code to the hybrid CPU/GPU environment and we immediately observed 5x acceleration.
This is a slam dunk ROI in a field where a 2 person-year effort to optimize code might be considered a success at even a 50% observed benefit. But what’s not included are the decades of prior investment that Mike and I have already made in the field. Our collective experience solving seemingly unrelated problems contributed to our success. Impact equations need to include intangibles such as the long-term benefit of cultivating a competitive workforce, or the opportunity cost of not staying current. These effects are cumulative, and once you get off the cutting edge, it becomes harder to get back in.
HPCwire: On a personal note, can you talk about your personal life? Your family, background, any hobbies?
Masako Yamada: I led an urban lifestyle until I moved to upstate New York in my late 20’s to work for GE. Home repair, yard maintenance and gardening were not a regular part of my upbringing. I never even had a car. To this day, I’m an avid “indoors-person” and my hobbies are piano, reading, writing, cooking, crafts and going to restaurants, concerts and museums, etc. I also enjoy teaching children as a volunteer tutor.
My colleagues at GE always encouraged me to take advantage of the beauty of the area, especially during the long winter months, but even after 10 years living upstate, I wasn’t inspired to change until the birth of my son in 2012. My husband and I now actively seek outdoor activities to enjoy with our son. Not only is cultivating a DIY spirit and an appreciation for nature a blessing, it’s equally important that our son feel comfortable in his own terrain, which is quite different from the environment in which my husband and I were raised.
HPCwire: One last question – is there anything about yourself that you can share that you think your colleagues would be surprised to learn?
Masako Yamada: People often don’t know where to place me because I have a Japanese name, a formal English vocabulary and an indeterminate American accent. I grew up in a very densely populated part of New Jersey, just a couple miles away from New York City. Even as a kindergartener, I walked to school by myself, often visiting the candy store on the way.
My friends and I organized our own games, rummage sales and talent shows, right in the middle of the street. We yelled “Car!” and ran onto the sidewalk whenever we heard an engine. No adults ever came by to coach our songs, applaud our skits or purchase our refreshments; in the evening, our moms yelled through the windows to tell us to come home. I realize times are different (my own life changed after I transferred to a Japanese school at age 11) but I hope my son will be able to experience the freedom to create and inhabit his own little world.