Intel Cluster OpenMP Workshop
April 21 , 2006
9:00 AM - 5:00 PM (CDT)
Room 1030 of the NCSA Building, 1205 W. Clark St., Urbana, IL [Driving
Directions]
Description:
Cluster OpenMP is an implementation of the OpenMP standard
that runs on
a cluster and provides a cost-effective way to use many processors
when running a single program. The OpenMP standard specifies a set of directives
for use in C, Fortran, and C++ that can be used to run programs in parallel
on shared memory multiprocessors. Cluster OpenMP extends this programming
model to non-shared memory machines interconnected by
a network, such as
Ethernet or Infiniband. A simple extension to the OpenMP standard (the sharable
directive) may be required in order to run a program using Cluster OpenMP.
This workshop is one of a series of workshops being offered by Intel to
teach programmers how to apply Cluster OpenMP to their own codes. Some
familiarity with OpenMP and parallel programming is desirable, although not
required. Attendees are urged to bring their own codes so they may
port them to Cluster OpenMP.
The morning session consists of training lectures on OpenMP and Cluster
OpenMP, followed by some hands-on laboratory exercises. The afternoon
session provides time for attendees to port their own codes, or to work with
codes provided by Intel. Access to an Intel cluster and workstations
is provided. The one-day workshop is free of charge and lunch is provided.
Agenda:
9:00-9:05 Welcome
9:05-9:35 Overview of OpenMP
9:35-10:35 Introduction to Cluster OpenMP
10:35-10:45 Break
10:45-12:00 Hands-on exercises
12:00-1:00 Lunch (provided)
1:00-1:15 Afternoon introduction
and planning
1:15-4:00 Code porting and/or
large exercises (breaks as needed)
4:00-4:30 Wrapup, instructors
available until 5:00
Target Audience:
Programmers interested in how to run their programs
using Cluster OpenMP.
Cost:
No charge
Registration:
Registration is closed.
Point of Contact:
For questions regarding this training event, contact the NCSA Training Group.