Computer Engineering Seminar

Exploiting Parallel Execution Models in OS Design – A Report on IBM Cyclops-64 System Software Model and Design

Guang GaoProfessorUniversity of Delaware

Operating Systems have been considered as a corner stone of modern computing systems, virtually powering all general-purpose computers today. The basic functions of contemporary operating systems are based on fundamental research in 1960s or 70s where these target computers are developed following a sequential execution model (so-called von Neumann model) developed in 1940s.

In this talk, we review the role of classical OS in computing systems and the challenges we are facing today/future that the target computers most certainly will be based on parallel computer architectures from bottom (hardware chip-level multiprocessors) up. We argue that a family of sound parallel program execution models are essential to exploit the many levels of parallelism both in hardware and software from both the architecture and system software perspective. We anticipate the arrival of a new era of interesting and fundamental research of OS design – that will be closely connected to the research of new parallel execution and architecture models and their compilation (both static and dynamic) technology. We illustrate our vision based on the design and implementation of the FIRST (Flexible Integrated Runtime Technology) based OS design for the IBM Cyclops-64 many-core chip based high-end computing systems.

Guang R. Gao received both his Masters and PhD degree in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology, in 1982 and 1988, respectively. He is the first from People’s Republic of China (mainland) to receive such computer science graduate degrees from MIT.

Dr. Gao is currently an Endowed Distinguished Professor of Electrical and Computer Engineering at University of Delaware, USA where he has founded and directed the Computer Architecture and Parallel Systems Lab (CAPSL).

Dr. Gao's main research interest is in high-performance/parallel computer systems: architectures, execution and programming models, compilers and system software (including compilers, OS and runtime). He has devoted much of his time in researching scalable parallel program execution and architecture models that can serve as a basis for high-end parallel supercomputers. Recently, Dr. Gao has been elected as a Fellow of both ACM and IEEE to recognize his contributions to parallel computer architecture and compilation technology.

Dr. Gao has led numerous research programs in parallel computing architectures and software systems sponsored by US NSF, DARPA, DOE, DOD and Canadian government agencies, and private organizations such IBM, Intel, HP, etc.. He has published widely in reputable international journals and conferences. He has been a co-founder of influential and prominent international conferences (such as PACT, CASES). He actively participated in many international conferences/workshops/meeting — as a member of organizing/steering/program committees (or as chairs sometime), or as a keynote speaker, invited speaker or panelist He has served as on the editorial boards of several international journals.

Sponsored by