13 Exercises

• Exercise 1: Confirm the values of q computed for the above three algorithms.
• Exercise 2: A comparison of blocked implementations for the basic linear algebra subroutine and the matrix-multiplication implemention, previously provided.
• Exercise 3: Implement Cannon's algorithm on a serial machine in your favorite programming language and confirm that it works.
• Exercise 4: Confirm this timing analysis for Cannon's method.
• Exercise 5: A consideration of the cost involved in an algorithm and a slight variation of the algorithm.
• Exercise 6: The algorithm derivation to return two matrices to their original state.
• Exercise 7: Illustrate the block scattered layout of a matrix on a processor grid with blocks.
• Exercise 8: Verification of Cannon's algorithm when the matrices are stored in a scattered manner.
• Exercise 9: An illustration of an analog of algorithm for Cholesky, and the derivation of other analogs for some algorithms.
• Exercise 10: An illustration of the use of a netlib routine, the modification, and the consideration of the errors.