Lecture
This lecture covers the basics of MPI, including the distributed memory programming paradigm, point-to-point communications, collective communications, and synchronizations. It explains how to run multiple instances of a program, different types of communications in MPI, and the importance of matching sends with receives to avoid race conditions or deadlocks. The lecture also delves into blocking and non-blocking point-to-point communications, MPI barriers, wildcard receives, and collective communications like MPI_Bcast, MPI_Scatter, MPI_Gather, and MPI_Reduce. Additionally, it discusses receiving image parts in order, out-of-order, and with a collective, as well as the importance of avoiding collective calls within conditional clauses and customizing communicators and datatypes.