Description
The aim of the course:
- an introduction to data input and output in parallel
- an introduction to input and output using MPI-IO
Description:
This short course introduces users to efficient parallel data input and
output with the MPI parallel programming library. A number of short
presentations and group practicals will allow attendees to obtain hands
on experience on the day. Despite its apparent complexity, MPI-IO is
very straightforward model. It is well worth the effort of learning how
to use MPI-IO as it enables parallel applications to use a single file
format regardless of the number of processes being used. In other words,
the format of the parallel file is identical to that produced by a
serial program.
Prerequisites:
The course is designed for those familiar with the UNIX environment and
experience of programming in C or FORTRAN is essential. Previous
experience of MPI and parallel programming is also necessary, either
gained through experience or by attending the 'Introduction to Parallel
Programming and MPI for C and FORTRAN' course. Experience of compiling
and submitting parallel jobs on ARC1 is also necessary. ***IF YOU DO
HAVE A SPECIFIC PARALLEL DATA ISSUE, YOU WILL BENEFIT MOST FROM THIS
COURSE. PLEASE BRING DETAILS WITH YOU AS THE TUTOR CAN ADVISE ON
SPECIFICS THROUGHOUT THE DAY.***
Suitability:
The course would most benefit users familiar with existing MPI codes, or
users with C or FORTRAN single processor serial codes that they would
like to develop for the University's high performance computing 2000-
processor facility, ARC1. The course does NOT cover using ARC1 (covered
in a separate course), MPI programming (covered in a separate course) or
shared memory programming.
Course Format:
A half day course consisting of short tutorials followed by practical
sessions. Course materials will provided on the HPC wiki for open access
or beforehand so you can print your own and bring them along. Places on
the course will be limited to ensure that delegates receive personal
attention.
Course Provider:
The course is led by Dr Chris Wareing, a research fellow in the School
of Process, Environmental and Materials Engineering and member of the
ARC1 HPC support team. As an academic, he has used the MPI language to
develop a variety of serial codes for use on parallel HPC machines over
the last ten years.
Duration: half day.
Cost: Free of Charge
Suitability: Research staff and research postgraduate students.
Max places: 8
|
