Lecture 11

Learning objectives

After this class, you should be able to:

1. Create user defined data types in MPI and use it in communication operations.
2. Perform performance analysis of parallel code, including timing, computing speedup, and analyzing its scalability.

Reading assignment

1. Lecture 11 slides.
2. Lecture 11 example code.
3. Section 3.5-3.6.
4. Review sorting from data structures and algorithms courses.

Exercises and review questions

• Questions on current lecture's material
  1. Modify the code from the first review question of Lecture 10 as follows. Instead of obtaining its argument from the command line, process 0 will read it from a file and broadcast it to all other processes. Also, time the integration and reduction step of your code, as in the example from lecture 11. Each process should output its rank and its time taken. Process 0 should also output the value of the integral.

• Questions on next lecture's material
  1. Summarize the sequential sorting algorithms that you have studied and give their worst and average case time complexities.