Accelerating Scientific Applications with Reconfigurable Computing

• Speaker: Dr. Volodymyr Kindratenko
• Location: CH430
• Date: Monday, June 25
• Time: 11:00 am - 12:00 pm

Seminar Abstract

Reconfigurable computing has been viewed by some as the next
supercomputing paradigm that has a potential to achieve new levels of
application performance beyond of what the conventional processors can
offer in the near future.  In the past decade reconfigurable computing
has evolved to the point where it can substantially accelerate
computationally intensive floating-point scientific codes at the same
time as providing orders-of-magnitude improvements in power, size, and
cost over conventional systems.  High-performance reconfigurable
computers based on the combination of conventional processors and
field-programmable gate arrays (FPGAs) are now a readily available
commodity as the major HPC vendors, such as Cray and SGI, as well as a
number of newcomers, such as SRC Computers and Nallatech, offer viable
systems.  Even the CPU chipmakers, such as AMD and Intel, have been
looking at providing support for FPGAs as co-processors to the
conventional processing elements.

Since 2004, Innovative Systems Laboratory at the National Center for
Supercomputing Applications has been researching the use of
high-performance reconfigurable computing technology for accelerating
scientific computing applications.  The challenges in porting existing
applications to reconfigurable computers are not trivial: there is a
long and error-prone path from the software/hardware code partitioning
to an efficient FPGA code implementation.  In this presentation, I will
provide an overview of the reconfigurable computing technology and will
highlight some of the latest scientific codes, such as molecular
dynamics and cosmological data analysis, successfully implemented on
SRC-6 and SGI RC100 reconfigurable computing systems.  Issues such as
software/hardware code partitioning, code transformations, performance
analysis, and load-balancing will be discussed in the context of
existing applications.

Speaker Biography

Dr. Kindratenko is a Senior Research Scientist in the Innovative Systems
Laboratory (ISL) at the National Center for Supercomputing Applications
(NCSA), University of Illinois at Urbana-Champaign (UIUC).  He has
worked in the field of high-performance scientific computing for over
ten years.  He received the Ph.D. degree from the University of Antwerp,
Belgium, in 1997 and the M.S. degree from the State Pedagogical
University, Kirovograd, Ukraine, in 1993.  Dr. Kindratenko's research
interests include high-performance reconfigurable computing systems and
their use in scientific computing.  He has been working with application
scientists on implementing scientific codes on reconfigurable computing
platforms focusing on the use of high-level languages and algorithm
optimization techniques.  Dr. Kindratenko's work has been funded by NSF,
NASA, NCASSR/ONR, and NCSA Private Sector Program.  He has published
over 30 papers in refereed scientific journals and conference
proceedings and holds 4 US patents.  He is a Senior Member of IEEE and
ACM.