Department of Energy’s Exascale Computing Project (ECP)
Lori Diachin is the Deputy Director for the U.S. Department of Energy’s Exascale Computing Project (ECP). She is also serving as the Deputy Associate Director (DAD) for Science and Technology in the Computation Directorate at Lawrence Livermore National Laboratory.
She previously served as the program director for the HPC4Manufacturing and HPC4Materials programs in the DOE EERE and FE program offices. She was involved with DOE’s SciDAC program from 2001-2018, most recently serving as the FASTMath Institute Director. She was the Director for the Center for Applied Scientific Computing (CASC) for 6 years and the Acting Information Technology Department head at LLNL for two years.
Lori Diachin has over 25 years experience in applied mathematics research where her areas of expertise include mesh quality improvement, mesh component software, numerical methods, and parallel computing. She is the co-author of over 50 technical journal articles, book chapters, and conference proceeding articles in these areas. Before joining LLNL, Lori was a computer scientist at Argonne National Laboratory and a Member of the Technical Staff at Sandia National Laboratory.
HPCwire: Hi Lori, congrats on your new role and your many accomplishments. What excites you about working in the DOE and Exascale Computing Project?
Lori Diachin: In the DOE, I am excited about being a part of a mission organization, and the fact that what we are doing is important and applies to problems that really impact the nation. In addition, there are many different things to work on throughout the course of a career in the DOE and, for me, many different labs to work at. I’ve been at three of the DOE national labs and have had many different assignments over the years, which allows for continuous growth and career revitalization. Finally, the people I work with every day, such as the technical researchers, are really great.
For the ECP in particular what I’ve found really exciting is the fact that it’s one of the largest and most important computing projects that the DOE has sponsored. We’re focused on developing a capable exascale computing ecosystem comprising application development, software technologies, and advanced hardware design. What we are doing could have a tremendous impact on our nation; both from the perspective of the applications we are developing and the investments we are making to keep the high performance computing ecosystem healthy. In my current assignment, I work primarily work with the ECP leadership team many of whom I’ve known for decades and they are outstanding. They’re dedicated to what we’re doing and all very capable; it’s a great team to work with. Doug Kothe, the ECP project director, has created a very collaborative dynamic that’s open and inclusive, and I really value that team environment. I also really enjoy learning new things, including the broad technical scope of ECP and the intricacies of managing a such a large, diverse research, development, and deployment project under formal DOE project management guidelines.
HPCwire: What elements of the Exascale Computing Project are you focused on and what do you see as (some of) ECP’s unique strengths?
I have only been with ECP for about 6 months so I’m pretty new to the effort and so initially I’ve been very focused on learning. Learning about the technical scope across all of the applications, learning more about the software technologies that are being developed, and learning about formal project management practices in the DOE. Currently, I’m settling in and I’m focused on a lot of different aspects of project management in partnership with Doug Kothe. One example would be integration across different ECP program elements. There are approximately 1,000 researchers at 16 different laboratories working on over 100 different subprojects in the ECP, tracking the integration that’s happening among all of those efforts is an interesting challenge. Other recent activities include budget planning, improving our monthly reporting processes, and preparing for our final design review, which is coming up this summer. Because Doug is very inclusive, he brings me in on many different aspects of the project, and it’s really been a lot of fun.
In terms of unique strengths, as far as I know, the ECP is the largest computational science, computer science, and math project in the DOE designed to develop an integrated set of technologies for a capable exascale ecosystem. There are 25 application teams focused on delivering science challenge problems that they couldn’t begin to tackle today. This will enable a lot of new capabilities that are important for the nation in the areas of energy, science, economic security, and national security. On the software side there are some unique opportunities enabled by the ECP because, for the first time, we have such a large integrated team that can all pull in one direction. For example, we have the ability to look at DOE-developed software and think about making it more easily accessible to end users. This is done through the definition of community policies that enable interoperability, along with the creation of software development tool kits and continuous integration practices that allow easier, turnkey access to numerical libraries, developer productivity tools, and data and visualization analysis techniques. This is something that we haven’t had much critical mass around in previous projects, and the scale of this effort is really unique.
HPCwire: What inspired you to pursue a career in STEM and what advice would you give to young people wishing to follow in your footsteps?
My father was a professor of math education, so growing up there was a lot of emphasis on science and math. This led me to get my BA in math. I learned a fair amount of theory with that degree, but I didn’t know the best way to apply it. As a result, I decided to continue my education in graduate school and get my degree in applied mathematics. Graduate school is where I first fell in love with the idea of numerical methods and how they can be used to solve really challenging physics problems by breaking them down into smaller pieces that are easier to solve on high performance computers. This is what I ended up focusing on for my dissertation, for my post doc, and for my first several years at Argonne National Lab (ANL).
Early in my career at ANL, I realized that while I enjoyed the technical work, I had an even greater passion for the management of science. So after about 10 years on the technical track, I started to focus more on project, line, and program management. That’s what I’ve been doing the last 20 or so years; increasing my responsibilities in those areas where I believe my skills are probably best suited.
When I first expressed interest in pursuing a career in scientific management, I was given the advice to establish my credentials as a scientist first. I gradually took on more and more project responsibilities, larger projects, and multi-institutional teams. I took on line management responsibilities here at Lawrence Livermore National Lab (LLNL) starting with 15 people and scaling to about 400 people. I’ve done many different things in program management at LLNL including managing our portfolio of internally directed research projects serving as the director of the DOE HPC4Manufacturing program.
Aside from following the advice given to me to establish their technical credentials before pursing management, I would advise young people to recognize and embrace their unique strengths. It’s the diversity in our skill sets that makes a team and organization stronger. You are responsible for your own career, so figure out what you are passionate about, what you are good at, and pursue the assignments and opportunities that highlight those skills. Another thing that really helped me was finding a great mentor who was doing a job I thought I would enjoy doing. I talked to that person extensively about how they got to where they were and how I could do something similar. His advice was, and still is, invaluable. I think great mentorship really makes a difference.
HPCwire: Generally speaking, what trends and/or technologies in high-performance computing do you see as particularly relevant for the next five years?
We are increasingly in an era where we have more specialized processors and where we’re using different processors together within a system. This is leading to much more heterogeneous computing, which has significant implications on our programming models, development tools, libraries, workflows, and applications.
One of the interesting aspects of this for me is that algorithms that were dismissed 20 or 30 years ago are being revisited because of the way architectures are changing. For example consider high order numerical discretization methods. When memory was free and flops were expensive, high order methods were seen as too expensive relative to the additional accuracy achieved and so researchers largely dismissed methods higher than second order as not worth the computational cost. Now that data movement is very expensive and the computing is “free”, researchers are finding that much higher-order methods return better arithmetic intensity, and therefore value, for the data motion that is needed. At LLNL, we’re seeing very good results with these methods and I expect they will gain even more traction in the years to come. This idea that the old algorithms are becoming new again is what I find interesting.
Another thing that must be mentioned to keep an eye on is explosion of data science. There are many groups and industries exploring data science techniques; what I’m most interested in is how we can leverage data science techniques to enhance and improve the scientific endeavors the DOE cares about, for example, physical science experiments or large-scale simulations that model physical processes.
HPCwire: Outside of the professional sphere, what can you tell us about yourself – personal life, family, background, hobbies, etc.? Is there anything about you your colleagues might be surprised to learn?
I’ve been married for 22 years, don’t have any kids but a lot of nieces and nephews who are great. What’s dominated my personal life most recently is a major remodel on a 1940’s home in downtown Livermore, which we are just finishing now. In terms of hobbies, I love to read, mostly fiction of all different genres. I like to draw and I’m currently trying to master watercolors, although I’m not very good at it yet. I love good food, good wine, and, increasingly, craft cocktails. I think often times people are surprised to learn I have an artistic side. I nearly majored in art in college, but I couldn’t quite figure out how that would pay the bills so I majored in math instead, and now, here I am.