Parallel Computing: Lecture 14

Learning objectives

After this class, you should be able to:

1. Explain the steps of the row-striped and checkerboard matrix-vector multiplication algorithms.
2. Analyze the time complexities of the above algorithms, and derive the iso-efficiency and scalability functions.
3. Develop and analyze matrix-vector multiplication algorithms based on other data-decompositions.
4. Develop and analyze algorithms for similar problems.

Reading assignment

1. Chapter 8, except columnwise-striped partitioning, and checkerboard partitioning where p is not a perfect square.
2. None. Prepare for midterm.

Exercises and review questions

• Questions on current lecture's material
  1. Exercise 8.3. You need not write a program. Instead, write the algorithm and analyze it as we did in class.
  2. Exercise 8.11. You need not write a program. Instead, write the algorithm and analyze it as we did in class.

• Questions on next lecture's material
  1. None.