University of Minnesota
University Relations
myU OneStop

Go to unit's home.

Home | Seminars and Symposia | Past seminars/symposia: Wednesday, November 7, 2012

DTC Leading Edge Seminar Series

High-Performance Computing in the Scientific Community: Partnering with Industry


Carlos P. Sosa
Cray, Inc.

Wednesday, November 7, 2012
3:30 p.m. reception
4:00 p.m. seminar

401/402 Walter Library

SosaIt is clear that the needs for high-performance computing in the scientific community have evolved. So have our technologies and systems. This community is rapidly moving towards an age of data-driven applications. This will continue to demand as much computational power as can reasonably be provided. Computational scientists in these fields have been early adopters of recent CPU-attached acceleration technologies (both GPUs and FPGAs), and Cray has a long history of delivering enhanced performance through acceleration starting with its early vector computers, today with its GPU-accelerated XK6 systems, and before the end of the decade with exascale capable systems. Recently Cray has introduced the uRiKA graph appliance built for Big Data relationship analytics. In this presentation, we review some of the efforts where Cray is changing the role of supercomputing. This includes High-Performance Computing, storage and Big Data. We also look at the importance of building a strong relationship with industry.


Carlos P. Sosa is an Applications Engineer in the Performance Group and Life Sciences Segment Manager at Cray Inc. Prior to joining Cray Inc, Carlos was a Senior Technical Staff Member in the High-Performance Computing Solutions Development Group at IBM, where he was the team lead of the Chemistry and Life Sciences group in the Blue Gene development group since 2006. For the past 20 years, his work focused on scientific applications with emphasis in Life Sciences and parallel programming. He received a Ph.D. degree in physical chemistry from Wayne State University working under the direction of Prof. Schlegel, one of the main developers of the Gaussian program. He completed his post-doctoral work at the Pacific Northwest National Laboratory. His research expertise falls broadly in the areas of molecular simulations, bioinformatics, and parallel computing. His research in parallel computing focuses on scaling applications in these areas to emerging hardware architectures such as the Cray flagships XK6 and XE6 supercomputers. He has led numerous strategic projects at IBM and Cray Research on addressing challenging scientific applications with cutting-edge high-performance computing technologies. He is the author of four IBM books as well as numerous publications in international journals and conferences. Currently, he holds an editorial board member of Current Drug Discovery Technologies and program committee member for HiCOMB 2013. He was also a visiting member of the IBM Academy of Technology. He is an Adjunct Professor at the Biomedical Informatics and Computational Biology, University of Minnesota Rochester. Prior to joining IBM Carlos worked at Cray Research in the Chemistry Group. There he helped parallelize the Gaussian94 program to the Cray T3E and later Carlos and co-workers used OpenMP to parallelize Gaussian on shared-memory machines.