Computer Science 400
Parallel Processing and High Performance Computing

Fall 2017, Siena College

Home     Syllabus     Schedule

Programming Project 2: Parallelizing Jacobi Iteration
Due: 11:59 PM, Monday, October 16, 2017

Your task is to parallelize the Jacobi iteration program using the SPMD model using message passing with MPI for communication.

You may use your own serial implementation as a starting point, or use my sample solution.

You may work individually or with a partner on this assignment.

Getting Set Up

You will receive an email with the link to follow to set up your GitHub repository jacobimpi-lab-yourgitname for this Programming Project. One member of the group should follow the link to set up the repository on GitHub, then that person should email the instructor with the other group members' GitHub usernames so they can be granted access. This will allow all members of the group to clone the repository and commit and push changes to the origin on GitHub. At least one group member should make a clone of the repository to begin work.

Parallelization of Jacobi Iteration

The parallelization will closely follow that in the MPI version of Conway's Game of Life. Distribute the computational domain among the processors by groups of adjacent rows. As with the Game of Life simulation, you may assume the number of processes evenly divides the number of grid rows.

Question 1: Find a problem size and a number of iterations that runs for about 10 minutes using a single process on noreaster. Run that same problem instance on 2, 4, 8, 16, and 32 processes and time each. (5 points)

For the questions below, assume a problem size of 512x512 grid cells.

Question 2: How many rows would be allocated in memory (a) on each process and (b) across all processes if you run on 1, 16, 64, and 256 processes? (4 points)

Question 3: How many messages are sent to pass a complete row from one process to another on each iteration when running on 16, 64, and 256 processes? (3 points)

Extra Credit Opportunities

For 5 bonus points, make your simulation work when the number of processes does not evenly divide the number of rows.

Submitting

Your submission requires that all required deliverables are committed and pushed to the master for your repository on GitHub.

Grading

This assignment is worth 70 points, which are distributed as follows:

> FeatureValueScore
Correct Makefile 1
Command-line parameters 2
Distributed grid allocation 10
Parallel Jacobi iteration 20
Stops based on iteration limit 2
Stops based on error tolerance 2
Print states/solution protected by #ifdefs 3
Print simulation stats 4
Documentation 9
Code Style 5
Lab Question 1 5
Lab Question 2 4
Lab Question 3 3
(Bonus) Works for any size grid and numprocs 5
Total 70