Parallel Computing: Lecture 15

Learning objectives

After this class, you should be able to:

1. Write MPI code using the following features: (i) MPI_Alltoall, (ii) MPI_Cart_coords, (iii) MPI_Cart_create, (iv) MPI_Cart_rank, (v) MPI_Comm_split, and (vi) MPI_Dims_create.

Reading assignment

1. Chapter 8.
2. Chapter 11, page 274.

Exercises and review questions

• Questions on current lecture's material
  1. Show the steps of an Alltoall operation on a hypercube with eight processors.
  2. (Programming) Write code where each processor has a variable called rank that stores its rank. Then, MPI_COMM_WORLD is split so that all even numbered processors are on one communicator, and all odd numbered processors are on another communicator. The even numbered processors perform a reduction on rank to output the sum of their ranks, and the odd numbered processors perform similar operations amongst themselves.

• Questions on next lecture's material
  1. (Post your solution on the discussion board) Give an example of two 3x3 matrices, and show their product.