Office Hours: 3pm+ MWF 718 SEO (by arrangement or email
preferable; sometimes there is and MCS committee meeting on Wed., then
see me at 4:30pm);
Catalog description:
Introduction to supercomputing on vector, parallel and massively
parallel processors;
architectural comparisons, parallel algorithms, vectorization
techniques, parallelization techniques, actual implementation on real
machines (Crays, Convex, Connection Machines and others).
Prerequisites:
MCS 471 Numerical Analysis
or MCS 571 Numerical Methods for Partial Differential Equations
or consent of the instructor. Graduate standing.
Semester Credit hours: 4
List of Topics --- Hours
Introduction to advanced scientific computing. --- 3 hours.
Comparison of serial, parallel and vector architectures. --- 3 hours.
Performance measures and models of performance. --- 3 hours.
Pure parallel algorithms and data dependencies. --- 3 hours.
Optimal code design. --- 3 hours.
Loop optimization by reformulation. --- 6 hours.
Code implementation on mini-supercomputers (eg, UIC Convex SPP1200).---4 hours.
Code implementation on vectorizing supercomputers (eg, Cray C90).---4 hours.
Code implementation on massively parallel processors (eg, Cray T3D).---4 hours.
Principle of locality of reference and cache storage. --- 4 hours.
Block decomposition methods. --- 3 hours.
Parallel programming packages. --- 5 hours.
Total. --- 45 hours.
Required Texts:
F. B. Hanson, "Local Supercomputing Training in Computational Science
and Engineering", February 1996