introduction -- speedup and efficiency
This course is part of the legacy of professor emiritus
Floyd Hanson;
see
MCS 572 Introduction to Supercomputing, Spring 2003.
Supercomputing is defined as the use of supercomputers and the goal
of this course is the learn how to use supercomputers, which translates
into parallel programming. The title of our textbook describes well
our activities. Our first exercise is to visit
top 500 supercomputing sites.
We stated Moore's law, defined flops, and related the clock speed
with flops. We defined speedup, efficiency, and gave an example of
superlinear speedup. We derived Amdahl's law, illustrated the law
and finished by stating that this law is too pessimistic and also
not realistic, promising to take scalability into account.
Bibliography
- Floyd Hanson: "Local supercomputing training in the computational
sciences using remote national centers". Future Generation Computer
Systems, vol 19, issue 8, pages 1135-1347, 2003.