From Code to Performance

Introduction to Git and Github

Learning Objective

After completing this guide, you will be able to:

• Explain CPU/node/interconnect architecture and why it matters for performance.
• Choose when to use MPI, OpenMP, or a hybrid approach.
• Apply performance-minded code patterns (blocking/tiling, non-blocking MPI, OpenMP reductions/tasks/SIMD).
• Measure, profile, and reason , about performance (strong/weak scaling, roofline model, Amdahl's Law, Gustafson's Law).
• Tune runtime and compiler flags, to improve performance.
• Debug common correctness and performance issues (race condition, false sharing, imbalance).