Visualize Numerical Integration

Libby Shoop and Ivana Marincic, Macalester College

Initial Publication Date: July 22, 2014

Summary

This is an activity with working code supplied that enables students to see how various forms of the data decomposition pattern map processing units to computations. The problem we use is numerical integration: approximation of the integral over an interval using the sum of areas of rectangles. Students run examples in OpenMP (threads) and/or MPI (processes) and a depiction of which processing units (either processes or threads) computed each rectangle is played back slow enough to see how the computation took place.

Share your modifications and improvements to this activity through the Community Contribution Tool »

Learning Goals

Students will be able to describe the types of data decomposition patterns that can be used in

distributed message passing situations using MPI,
shared-memory situations using OpenMP, and
hybrid cases, using both message passing and OpenMP

Context for Use

At a minimum, you will need a multicore machine with an OpenMP C compiler, such as gcc. This machine and any client from which the student would log in remotely will also need X11. If you want the MPI examples, you will need to compile and run the supplied code on a single machine or cluster with MPICH2 installed. This is an activity students can step through during class or lab. It is also quite suitable as an instructor demonstration in class.