Overview

Goals

• What is parallel programming?
• Why parallel programming?
• Performance is important
• How to parallelize a program?
• Tools
• Principles and best practices
• Strategy for scientific software development

Background knowledge

• The working of a modern processor
• CPU architecture and hierarchical memory architecture
  • A short introduction: Memory location matters for performance
  • A very good talk about this topic, you need to see this: Scott Meyers on Cpu Caches and Why You Care
• Accelerators (GPU), increasingly important topic, but we cannot treat everything in this course.
• The architecture of a supercomputer
• Nodes and cores, NUMA domains
• Interconnect