70154 - Fundamentals of Parallel Computing

Learning outcomes

At the end of this course, the student is able to design and implement parallel algorithms on shared-memory and distributed-memory architectures, and evaluate their performance. This course describes the main hardware platforms for parallel computing (multi-core processors, clusters of workstations, GP-GPUs, special purpose parallel machines). The concepts of speed-up, load balancing, workload distribution will be also described. Students will learn how to program some of the parallel architectures using standard tools (e.g., MPI and/or OpenMP). Applications in the area of numeric analysis and data mining will be briefly considered.

Course contents

• Introduction to parallel architectures: clusters, shared-memory multiprocessors, GPGPUs
• Parallel programming paradigms
• Performance evaluation of parallel programs
• An introduction to parallel algorithms (sorting, searching, numerical algorithms)
• Tools and languages for parallel programming
• Parallel programming with MPI and OpenMP

Requirements: during the course we will use of the C programming language; therefore it is required that students are familiar with imperative programming using one of C, C++ or Java, and are able to implement simple algorithms. Students of Informatica per il Management should have already passed the courses of Programmazione Internet, Laboratorio di Programmazione Internet and Algoritmi e Strutture Dati (and should have passed the final exams).

Readings/Bibliography

• Blaise Barney, Introduction to Parallel Computing, Lawrence Livermore National Laboratory
• I. Foster. Designing and Building Parallel Programs. Addison-Wesley, 1995 (Available online).

Additional learning material and textbooks will be proposed during the lectures.

Teaching methods

Lectures and programming exercises

Assessment methods

Written exam or programming project

Teaching tools

Lecture slides, source code of programming examples.

Links to further information

http://www.moreno.marzolla.name/teaching/ParComp2012/

Office hours

See the website of Moreno Marzolla