HPC Exercise 4:
MPI Collective Communication, Parallel I/O, and Combining Patterns
Objectives
After completing this exercise, you will be able to:
-
Use the Data Decomposition.
Master-Server, Parallel Loop,
Message-Passing, Collective Communication,
and Parallel I/O
patterns to solve a problem.
-
Use compiler optimization to speed up a sequential computation.
-
Use parallelism to speed up a sequential computation.
-
Analyze a computation to determine its performance bottlenecks.
-
Compute the speedup and efficiency
of a parallel computation.
Procedure
This week's exercise consists of 3 parts, that you should complete in order:
-
Lab exercise Part 1: Performing Sequential I/O Faster
-
Lab exercise Part 2: Collective Communication and Parallel I/O
-
This week's homework project
CS >
374 >
Exercises >
04
This page maintained by
Joel Adams.