High Performance Fortran in Practice
Charles Koelbel, Rice University
High Performance Fortran (HPF) was defined in 1993 to provide a portable
syntax for expressing data-parallel computations in Fortran.
appearance of the High Performance Fortran Language Specification
(available as an issue of Scientific Programming and by
there has been great interest
in HPF as a language for efficient parallel computation.
The purpose of this tutorial is two-fold:
- To introduce programmers to the most important features of HPF,
including features inherited from Fortran 90, the data parallel
FORALL statement and
and data mapping by
- To illustrate how these features can be used in practice on
algorithms for scientific computation such as LU decomposition and
the conjugate gradient method.
The intended audience of this tutorial is researchers and practitioners
who are interested in applying data-parallel computation to scientific
programs. Attendees should have some knowledge of FORTRAN 77 (or a
similar imperative sequential programming language); a basic
knowledge of scientific computation and/or parallelism is also useful, but
The tutorial can be accessed in several ways:
Choose the format that best fits your temperment. You can generally get
from any form to any other, if you follow enough links.
- Introduction to Data Parallelism
- Fortran 90 Features
- Data Parallel Features
- Data Mapping Features
- Parallel Programming in HPF
- HPF Version 2.0
Additional information about HPF is available at the following places:
This tutorial is available in Postscript form by
(Due to size, the tutorial appears in several parts.)
This HTML document summarizes a full-day tutorial on "High Performance
Fortran in Practice". The tutorial has been given (in various forms) at
the Seventh SIAM Conference on Parallel Processing for Scientific Computing (San
Francisco, CA, February 1995),
Supercomputer '95 (Mannheim, Germany, June 1995),
Supercomputing '95 (San Diego, CA, December 1995).
Last modified: April 30, 1996