Eric Roman

Eric Roman is a Computer Systems Engineer and Manager at the National Energy Research Scientific Computing Center (NERSC), part of Lawrence Berkeley National Laboratory. He has been with Berkeley Lab since 1999 and currently works in the HPC Technology Department within the Computing Sciences division.

Dr. Roman earned his PhD in physics from the University of California, Berkeley in 2010, with a dissertation entitled "Orientation Dependence of the Anomalous Hall Effect in 3D Ferromagnets." His doctoral research involved ab initio simulations of nonlinear optical properties of semiconductors, spin transport in metals, and the anomalous Hall effect.

Roman's primary research focus is on operating systems for high performance computing, with significant contributions to Berkeley Lab's Checkpoint/Restart (BLCR) technology since 2001. His work spans several key areas:

• Development of multithreaded checkpoints and restarts
• Implementation of file and pipe support in BLCR
• On-the-fly compression of checkpoint files
• Direct I/O capabilities for HPC systems
• Integration with batch systems like Torque
• Optimization of file I/O operations

His publication record shows consistent contributions to HPC systems research over 15+ years, with recent work focusing on resilience techniques, failure prediction, and system optimization. Roman has collaborated with researchers from multiple institutions, advancing parallel computing, fault tolerance, and system-level technologies that enable scientific discovery at scale. His highly cited works on live migration and checkpoint/restart frameworks demonstrate significant impact in the field.

Roman has been actively involved in the HPC community, leading Linux kernel seminars and organizing projects like "High-End Computing with K42" under the FastOS initiative. He continues to collaborate with the Berkeley ParLab on cutting-edge research in high-performance computing systems.