Skip to Main Content U.S. Department of Energy
High Performance Computing

At PNNL, high-performance computing encompasses multiple research areas that affect both computer and domain science. Our scientists and engineers enable high-performance computing for solving scientific problems by developing and implementing high-level programming; generating algorithms that manage computational complexity; and creating models, high-speed networks, and tools to provide problem-solving environments that enhance system productivity and` increase the use and availability of applications to non-specialists.


Advanced Architectures

Researchers focus on developing simulations and prototypes of various advanced computing architectures to facilitate system designs that result in balanced, efficient networks and improved operations.

Modeling and Simulation

Modeling and simulation techniques, including analytical modeling, emulation and simulation, statistical and stochastic, and other integrated approaches, are providing a quantitative understanding of the performance, power, and reliability of systems and applications.

Scalable Machine Learning

Machine learning techniques take inputs from sensors or images and configures them to “learn” about things. For example, the box that appears around someone as a portrait is taken with a smartphone camera is an element of machine learning that recognizes “this is a face.” PNNL is developing new machine learning methods and applications to understand the configurations that make them operate better and link them to run on some of the world’s largest computers.

System Software and Applications

This research, which focuses on tools and methods for efficient execution of applications on extreme-scale systems, encompasses all aspects of the software stack, including operating systems, low-level and application-facing runtime systems, programming models, and applications.

Testbeds and Tools

Diverse, practical testbeds and tools that employ state-of-the-art instrumentation. These testbeds include component technologies and scalability platforms that can be examined empirically and combined with accurate predictive performance modeling and simulation to facilitate co-design and examine the impact of technologies for possible future systems.

High Performance Computing


Seminar Series