CSci6702 Parallel Computing

1
CSci6702Parallel Computing

• Andrew Rau-Chaplin
• arc_at_cs.dal.ca
• www.cs.dal.ca/arc

2
Course Objectives

• Understand Parallel
• Architectures
• Systems
• Algorithms
• Learn how to
• Design efficient parallel algorithms, and
• Implement them on parallel machines
• Experience the Research Process

3
Official Outline

• This course explores various aspects of parallel
  computing including parallel architectures,
  algorithms, systems, programming languages and
  implementation issues.
• The focus is on solving real problems on existing
  parallel machines.
• Student will be expected to read recent research
  publications, and participate in significant
  parallel implementation projects.

4
Unofficial Outline

• Clusters, stacks of commodity boxes, have taken
  over the world of parallel computing.
• Learn how to take a stack of Pentium boxes and a
  Linux  CD and build your own parallel machine.
  Now for the harder trick - learn how to design
  algorithms and systems that can take advantage of
  all those cycles.
• Be ready to read algorithms papers, write tricky
  C and MPI code and learn to love your parallel
  debugger and profiler.

5
Resources

• Course web page
• www.cs.dal.ca/arc/teaching/CSci6702
• All notes, readings, assignments
• Parallel Machines
• Available from ACENET

6
Assignments

• Assignments (40)
• 1) Designing Parallel Algorithms
• 2) Programming
• Clusters in MPI
• SMP machines in OpenMP
• Seminar (20)
• Seminar on recent research topic. Groups of 2.
  Topics to be approved.

7
Research Project (40)

• Select your own topic
• Algorithms, Systems, or Application topics
• Components Literature review, some research or
  programming work, final paper, presentation
• Main Deliverable Conference style paper plus
  short in-class talk
• Groups of 1-2
• See project description page
• Mini Conference
• Note Papers must be done in LaTex!