Lecture 3 hours.
Prerequisites: 665.
This is a first course in parallel architecture, with an emphasis on the
description and evaluation of commercially available machines.
Topics to be covered include:
parallelization and performance metrics,
scalability and the ``laws'' of Amdahl and Gustavson, computational
similarity, models of computation,
parallelization paradigms, network characteristics and topology,
communication calculus and templates,
pipelining and parallelism, processor types,
memory hierarchy, cache coherence protocols, latency-hiding
mechanisms, routing algorithms, and languages and
libraries to support parallel architecture.