The goal of this training is to provide knowledge on major topics to achieve program performance:

• How to optimize a program to obtain the best performance, and
• Parallel programming - how to port a program to a parallel computer to reduce the execution time by dividing the work among several threads/processes.

Topics Covered

• Parallel Computer Architectures
• Compilers and Monitoring Tools
• Program Analysis and Optimization
• Performance Tuning
• Shared Memory Parallel Programming using OpenMP
• Parallel Programming in Distributed Memory using MPI

Objectives
Upon completion of this course, the student should be able to:

• Understand basic parallel computer architectures
• Use system tools to identify performance problems.
• Evaluate program performance
• Better use compiler/mathematical libraries capabilities.
• Understand basic optimization techniques
• Recognize the advantages and pitfalls of parallel processing.
• Write multithreaded programs using OpenMP
• Write multiprocessor programs using Message Passing Interface - MPI

Target Audience
This course is intended to programmers who work currently with C and Fortran, want to be able to recognize performance problems in their application, and to improve the execution time by program optimization and parallel execution.

Prerequisites
Students should have C or Fortran programming experience, and know how to perform basic Linux/Unix commands (cp, ls, make etc)