Lecture 16

Learning objectives

After this class, you should be able to:

1. Parallelize matrix-vector multiplication using pthreads.
2. Explain the idea of race conditions and critical sections.
3. Explain some disadvantages of the busy-waiting solution provided in the text book for mutual exclusions.
4. Use pthreads locks to implement mutual exclusions.

Reading assignment

1. Section 4.3 - 4.6.
2. Lecture 16 slides.
3. Lecture 16 example codes.

Exercises and review questions

• Questions on current lecture's material
  1. Rewrite the pi example from the previous homework so that all threads directly update the partial sum, using pthreads locks to avoid race conditions.
  2. (Post on the discussion board) Modify the matrix-vector multiplication code from the book so that it has two types of parallelization: cyclic and block. Give the time taken by each version.

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