NEW to Berkeley Lab Computing Sciences Team
May 26, 2009
 |
| ESnet's Eric Pouyoul |
Eric Pouyoul, Computer Scientist for ESnet's Advanced Technologies Group
As a computer scientist on ESnet's Advanced Technologies Group, Eric Pouyoul will be designing and developing software to manage the facility's On-Demand Circuits and Advance Reservation System (OSCARS) and PerfSONAR.
OSCARS is a network service that allows users to reserve bandwidth for latency-sensitive (time-sensitive) high-speed data transfers. PerfSONAR is a network performance monitoring and diagnostic system that enables network engineers to identify bottlenecks, which allow them to make relatively small tweaks to gain significant speedups.
A computer scientist by training, Pouyoul notes that he has developed software for extremely large high performance computers, or very small computer chips that are embedded on credit cards for security purposes.
“I am an extremist when it comes to computers, I like the really big machines or extremely small chips, and avoid anything in between,” jokes Pouyoul.
Although he has worked with high-performance networking in local area settings, Pouyoul admits that he is excited to contribute to networking across wide-area settings, which not only connect supercomputers but supercomputing centers.
“ESnet is a pioneer in this arena, and I am excited to join their team and learn from the experts,” he says.
Originally from Paris, France, Pouyoul began working in the United States in the 1990s when his employer, Chorus Systems, sent him to Northern California to act as an onsite consultant for their client, Unisys. He loved the area so much that he decided to stay.
Pouyoul currently lives in San Francisco, and likes to camp, hike, and dabble in photography in his spare time.
 |
| CRD's Mark Howison |
Mark Howison, Computer Systems Engineer in Visualization and Analytics Group
This month, Mark Howison went from student assistant to full-time computer systems engineer in the Berkeley Lab's Visualization and Analytics Group. In this role he will support NERSC users' analysis and visualization needs, as well as conduct basic research into scientific data visualization.
“As a student assistant, I worked primarily on improving parallel I/O performance on Franklin. Parallel I/O is an important step in the visualization pipeline for large datasets, and I am excited to continue this work at NERSC,” says Howison.
As a full-fledged member of the Visualization Group, Howison will continue this work by helping to optimize a new release of HDF5, a versatile I/O framework for large scientific datasets, specifically for Franklin.
“Horst Simon's graduate course on parallel computing at the University of California, Berkeley is what drew me into the HPC field,” says Howison, who recently earned a master's degree in Computer Science from UC Berkeley. He came to computer science by way of mathematics.
“I was always somewhat unsatisfied with studying math for maths sake, and seeing how HPC uses mathematics and computer science to solve applied and scientific problems was invigorating,” Howison adds.
Originally from Portland, Maine, Howison completed his undergraduate degree at Brown University. He likes to spend his spare time outdoors, hiking, camping, skiing, and “building things.” He cites a summer spent constructing a 20-foot wooden sailboat with his grandfather as one of his crowning achievements.
