Usage Notes for
                     IMSL(R) Fortran Numerical Library
                             Version 7.0.0


This document contains usage notes for the IMSL Fortran Numerical 
Library for AMD64 systems.

The IMSL Fortran Numerical Library Version 7.0.0 is thread-safe.

The IMSL Fortran Numerical Library is comprised of the products IMSL
MATH/LIBRARY, IMSL MATH/LIBRARY Special Functions and IMSL STAT/LIBRARY.

In this document IMSL MATH/LIBRARY, IMSL MATH/LIBRARY Special Functions and
IMSL STAT/LIBRARY will be referred to collectively as the IMSL Fortran 
Numerical Library.


===============================<  Contents  >===============================

This file has the following parts:

      Part    Contents
      ----    --------
      1       Introduction
      2       Supported Platform
      3       Miscellaneous Information
                  - Known Problems, Limitations, and Differences
      4       Performance Enhancements
                  -  Multiple CPU Usage in a Shared Memory Parallel Environment
      5       Customer Support


=========================<  Part 1: Introduction  >============================

This file contains information on the installation and usage of the IMSL 
Fortran Numerical Library which is specific to the environment documented below.

In addition to this README file there is a primary README file that
documents information which applies to all environments and a RELEASE_NOTES
file which documents changes made since the last release of the IMSL Fortran 
Numerical Library.  These files are located in this directory.


=========================<  Part 2: Supported Platform >===============

Below is the supported environment.


               IMSL Fortran Numerical Library
               SUSE Linux Enterprise Server 11 (x86_64)
               Linux Kernel: 2.6.27.19-5-default
               pgf90 10.2-0 64-bit target on x86-64 Linux -tp k8-64e
               LAPACK:    ACML (bundled with compiler)
               BLAS:      ACML (bundled with compiler)
               MPI:       bundled with compiler
               ScaLAPACK: bundled with compiler
               BLACS:     bundled with compiler
               Compiler Options:
                  shared library: -O -mp -tp k8-64e -Kieee -fpic
                  static library: -O -mp -tp k8-64e -Kieee

  
===================<  Part 3: Miscellaneous Information  >====================

     
- Known Problems, Limitations, and Differences

  o The IMSL routines arpack_symmetric, arpack_svd, arpack_nonsymmetric,
    and arpack_complex require that class objects be supported by the
    compiler.  Since this environment does not support class objects
    these four routines are not provided.

  o NVIDIA hardware is not supported.

  o The high performance option for the FFTs is not supported.

 
===================<  Part 4: Performance Enhancements  >====================

-  Multiple CPU Usage in a Shared Memory Parallel Environment

   SMP support is obtained via OpenMP directives and leveraging AMD Core
   Math Library(ACML).

   Because of the overhead required by the system to spawn threads, it is not
   always advantageous to use SMP capabilities.  In some cases where the
   amount of work that can be done in parallel is small when compared to the
   overhead, the application can actually take a performance hit due to the
   overhead for parallelism. If performance is a concern, benchmarking of the 
   application should be done using different LINK options.

   Once your application is linked, set the environment variables 
   OMP_NUM_THREADS to the number of CPUs you want your application to use.

=========================<  Part 5: Customer Support >===============

If you have problems installing/using any product, contact IMSL
Customer Support. The Customer Support group researches and answers 
your questions about all IMSL products.  Contact information can be 
found in the file README located in this directory.