HPC Research and Education Suite 100, 35764 Congress Road Farmington Hills, Michigan 48335-1222 Phone: (248) 477-7018 Fax: (248) 477-3129 Email: m.ginsberg@ieee.org
Myron has over twenty-five years experience in high performance computing (HPC) in private industry, academia, and government research labs. His research interests are in large-scale scientific / engineering computations and in benchmarking methodology for assessing the performance limitations of modestly and massively parallel computing systems. He is the first ACM Fellow from the world automotive industry and was cited for his "pioneering and sustained contributions to supercomputing research and its application to the automotive industry in addition to distinguished teaching and service in HPC." He has served as an distinguished national lecturer for ACM, SIAM, IEEE, ASME, SAE, and Sigma Xi. He holds a BA and MA in mathematics and a Ph.D. in computer science. Dr. Ginsberg is currently an HPC consultant to the automotive industry, having previously been a staff research scientist at General Motors Research.
This presentation will offer some pragmatic guidelines to cope with career problems. Topics include: continuing education strategies; creating and utilizing a rolodex; necessity of lifetime professional activity; strategies to prevent and/or deal with periods of unemployment and career changes; vita and resume suggestions; the politics of hiring; becoming a chameleon for professional longevity; acquiring multidisciplinary experience.
The proliferation of high-end workstations, machine clusters, and/or MPPs has made it very difficult to determine which machine is "best" for a specific application. Simply testing one or more kernels on a new computer is insufficient because slight program modifications can produce large variations in performance and thus could lead a user to erroneous conclusions about the behavior of the entire application. Industrial problems are far too large and complex to be easily or quickly benchmarked. Furthermore, both dedicated standalone performance and throughput tests must be performed to determine effectiveness in an industrial setting. This presentation will offer some guidelines to assist the user.
Most industrial computing positions require multidisciplinary skills which very few undergrads or grad students directly acquire during their academic career. There are about 30 computational science programs in the U.S. at present. This talk defines some of the essential skills that students should be directly exposed to in such a program. Topics include: direct lab experience in multidisciplinary projects; learning to be an intelligent novice in someone else's discipline; how to communicate with people in other disciplines; recognizing the transportability of problem-solving techniques; obtaining industrial manpower and financial support for such a program.
The U.S. government is trying to motivate the development of innovative computer architectures to satisfy the current and future industrial, government, and academic needs for very large-scale computations. This lecture deals with the roadblocks to creating such machines, examples of some current efforts, and real-world applications that could benefit from the use of such computers.
Additional presentations are available either related to the automotive industry and/or combinations of the above topics to meet specific audience needs or interests. Also during the lecture visit, it is highly desirable to set aside time to meet both with faculty and students to discuss technical and educational issues as well any professional or career concerns.