High Performance Parallel Computing Boot Camp

Registration for the Mid-Atlantic Crossroads (MAX) and University of Maryland joint Summer HPC Boot Camp is now open. This boot camp is a cooperative venture between the University of Maryland’s Office of Information Technology and the Mid-Atlantic Crossroads and is open to participants from all academic institutions and federal agencies in our region.

Date: August 16 - August 20, 2010

Location: University of Maryland, College Park
Cost: $50 for attendees, and you will need to register
Attendance is Limited to 30 participants
Sponsored by the University of Maryland’s Office of Information Technology and MAX
Instructor: Aaron Bloomfield, University of Virginia Computer Science Department
Payment: Checks, vouchers, JV can be sent to:
Maddie Parra (mparra@umd.edu)
1313 Computer and Space Sciences Building
College Park, MD 20742

Response to this Bootcamp has now exceeded the space for others to attend. Further registration is now closed for this event. Another will be planned for the Spring 2011.

Outcomes

Targeted at graduate students, staff, and faculty members with computational science and engineering problems that demand high performance, this course will introduce participants to the basics of high-performance parallel computing. Successfully completing the course, attendees will know how to:

• optimize sequential applications
• understand the basics of parallel computing
• write basic MPI and OpenMP applications

Attendees will use queuing systems such as PBS and existing high-end resources at the University of Maryland during the course.

Who Should Attend?

Faculty, graduate students, and research staff from Mid-Atlantic region universities and national laboratories experiencing any of the following issues:

• Computational science or engineering problems requiring high performance
• Problems thought to be too computationally difficult to tackle
• A desire for their programs to run faster and complete sooner

Topics to Be Covered

• Performance measurement and optimization
• Parallel computing concepts
• Tightly coupled MPPs
• Limits to parallelization
• Distributed memory computing: MPI (Message Passing Interface)
  * Simple stencil problems (e.g., explicit methods)
  * More complex irregular structures
  
• Shared memory computing: OpenMP
  * Cache Coherence
  
• Parallel program design

Prerequisites

Participants should be experienced with one or more of the following programming languages:

• C
• C++
• Fortran
• UNIX basics such as editing, compiling, the file system, and simple scripts

Format

The course will be composed of morning lectures and afternoon hands-on computer exercises with multiple support staff present to assist participants. Morning and afternoon snacks will be provided.