Description
You are required to pick a problem of your choice to parallelize.
The problem should be one that requires some amount of message
passing at each step, i.e. a loosely synchronous or synchronous
problem. Some examples, such as N-body Problem, have been given
in the lectures. You can choose one of the examples if you wish.
You are required to write an MPI+OpenMP/GPU program that can
run on multiple nodes of the cluster, using multiple cores per node, for
your selected problem.
Write up to 2000 words that outlines your problem and how you
achieved parallelism/high performance. Include tables and/or charts of
your own measurements that support your discussion.
Aaron Harwood and Lachlan Andrew (School of Computing and Information Systems The University of Melbourne) COMP90025 Parallel and Multicore Computing 2019 Semester II 2 / 4
Assessment
Project 2 is worth 20% of your total assessment. It is group work.
Assessment of the report (8/20) is based on the level of details and
presentation.
Assessment of the program (12/20) is based on your parallel
techniques employed/investigated.
Aaron Harwood and Lachlan Andrew (School of Computing and Information Systems The University of Melbourne) COMP90025 Parallel and Multicore Computing 2019 Semester II 3 / 4
Submission
Submit a PDF of your report (use PDF only, no other format will be
assessed) via LMS on or before Saturday 26th October. As well you
will need to submit your program. Instructions for doing this will be
given closer to the deadline.
Use 10pt font, single line spacing, 1 inch margins all around and
double column. Use appropriate headings and clearly label and refer to
tables/figures. Clearly put your names and login names at the top of
the report.
Aaron Harwood and Lachlan Andrew (School of Computing and Information Systems The University of Melbourne) COMP90025 Parallel and Multicore Computing 2019 Semester II 4 / 4