Introduction to High Performance Computing: Tools and Algorithms

Catalog description

This course will cover fundamental concepts of parallel computing: shared and distributed memory models; metrics for performance measuring; roof-line model for analysis of computational kernels, prediction and improving their performance on different processors; code optimization. We will analyze algorithms maximizing data reuse, and memory bandwidth utilization. Prior experience in coding is a plus. One course meeting will take place at IBM/Research, students will interact with experts in areas of HPC, visualization, social media and more. There will be bi-weekly homework assignments and a final project. Students are encouraged to suggest final project relevant to their research and level of expertise.