Identifying and tackling the numerous challenges for Exascale computing systems that will consist of millions of cores/nodes; not just homogeneous cores but cores of different types is a huge challenge. Accelerators go beyond just GPU(s) and Phi(s), opening up programmer’s Pandora’s Box. Is it even possible to maintain a single code base? There is a dire need to investigate the trade-off between performance and expressivity. Addressing issues of fault tolerance and power consumption are critical to the success of Exascale system. This talk will throw light on some of these issues.

Bio:
Sunita Chandrasekaran is currently a PostDoc in the Department of Computer Science at the University of Houston. Her area of research spans exploring programming models for multicore systems and accelerators, investigating programming requirements for complex, irregular scientific applications, estimating power & energy for scientific kernels using statistical methods. She received her Ph.D. degree in Computer Science Engineering from Nanyang Technological University (NTU), Singapore and B.E degree in Electrical & Electronics from Anna University, India.