Title: Programming Models for Heterogeneous Multi amd Manycores

Time：May 19, 2010, 4:00pm-5:30pm

Speaker: Dr. Rudi Eigenmann

Place：FIT 1-315

Abstract:

Multicores that include heterogeneous processors and distributed address spaces will likely find themselves among the emerging architectures. Heterogeneity will emerge because specialized cores will be able to accelerate specific program patterns. Consequently, the program will have to orchestrate the execution of different program sections on the most suitable cores. Furthermore, with large core counts (manycores), homogeneous, shared address spaces become less feasible. Programs will have to distribute their data and perform inter-core data transfers explicitly. The rising software complexity is evident.

In this talk, I will explore the possibility of providing a simple shared-memory model to the programmer, while performing the tasks of orchestrating heterogeneous cores and data distribution and transfer automatically. I will describe recent projects to develop compiler and runtime techniques that perform these tasks in the context of OpenMP shared-memory programs, the CUDA model for GPUs, and MPI for explicit message passing.

Biography:

Dr. Rudi Eigenmann

Rudolf Eigenmann is a Professor in the School of Electrical and Computer Engineering at Purdue University and serves as the Technical Director for High-Performance Computingof the Computing Research Institute. His research interests include optimizing compilers, programming methodologies and tools, performance evaluation for high-performance computers and cyberinfrastructures. He is currently serving as co-Principal Investigator of the Network for Earthquake Engineering Simulation (NEES) Operations project - a large effort involving 20 Universities.

Dr. Eigenmann has published his work in over 150 papers in international conferences, journals, and workshops. He received his Ph.D. in Electrical Engineering/Computer Science in 1988 from ETH Zurich, Switzerland. From 1988 to 1995 he worked as a research scientist at the Center for Supercomputing Research and Development, University of Illinois at Urbana Champaign, where he also served as the leader of the Center's Cedar Fortran compiler group. He is the recipient of a 1997 NSF CAREER award and serves on the editorial boards of the International Journal of Parallel Programming, the IEEE Transaction on Parallel and Distributed Systems Journal, and the IEEE Computing in Science and Engineering Magazine. He has served as the chairman of Computer Engineering at Purdue's School of ECE and as the chairman of the High-Performance Group of the Standard Performance Evaluation Corporation (SPEC). He has also been the general chair and program chair of such conferences as the ACM Symposium on Principles and Practice of Parallel Programming, the International Conference on Parallel Processing, the Workshop on Languages and Compilers for High-Performance Computing, and the Workshop on High-Level Interfaces for Parallel Systems. Address: Purdue University, School of Electrical and Computer Engineering, 465 Northwestern Avenue, West Lafayette, IN 47907, USA.

eigenman@purdue.edu

http://www.ece.purdue.edu/~eigenman