Some entries are incomplete. Last update in 5/02/1999.
How to use various programs
mpi
To compile and link {F77 | F90 | C |C++ } programs calling MPI routines (C++ programs cannot use C++ MPI Bindings) use {mpif77 | mpif90 | mpicc | mpiCC} as the compiler. They accept all options of {pgf77 | pgf90 | pgcc | pgCC} in addition to a few others. Check the respective man pages.
Example : mpif77 -O2 -tp p6 test.f -o test
To run programs using MPI routines use mpirun. The necessary option is
-np; it defines the number of processes. For the others consult the man
pages.
Example : mpirun -np 4 test
To view the man pages you can use mpiman.
queues
-
The queuing system was reconfigured as following:
-
-
queue |
Memory |
CPU time |
nice |
jobs/user/system |
|
|
|
|
|
large |
250 MB |
unlimited |
15 |
2 |
m-large |
900 MB |
unlimited |
15 |
1 |
interactive |
unlimited |
unlimited |
0 |
unlimited |
-
to submit a job:
> qsub -q queue
where queue: the queue name,
eg. fast, medium ...
batch file example:
cd /home/myname/programs/executables
program_in_the_above_directory.exe
to check the jobs:
> qstat -da
to delete a job:
> qdel -k ID.fire@machine
-
where:
ID : the ID shown in qstat
machine:
the machine where the job is running
-
eg. : qdel -k 19.fire@water
-
LessTif 0.87.0
-
A free Motif clone. Necessary to build Xmpi. Visit the Homepage
of Lesstif.
Xmpi
-
XMPI is a graphical user interface for running MPI programs, monitoring
MPI processes and messages, and viewing execution trace files. It exploits
the debugging capabilities of LAM, a parallel computing environment for
UNIX clusters. XMPI is constructed from the Motif widget set.
Further info
PGapack v1.0
-
PGAPack is a general-purpose, data-structure-neutral, parallel genetic
algorithm library being developed at Argonne National Laboratory.
Key features are:
-
Callable from Fortran or C.
-
Runs on uniprocessors, parallel computers, and workstation networks.
-
Binary-, integer-, real-, and character-valued native data types.
-
Object-oriented data structure neutral design.
-
Parameterized population replacement.
-
Multiple choices for selection, crossover, and mutation operators.
-
Easy integration of hill-climbing heuristics.
-
Easy-to-use interface for novice and application users.
-
Fully extensible to support custom operators and new data types.
-
Extensive debugging facilities.
-
A large set of example problems.
Further
info
Comments:
-
Details in the manual
-
to view the man pages: man 'routine-name'
-
In linking add: -lpga0
Parmetis v2.0 and metis v.4.0
-
Parallel Library for Unstructured Mesh (Re)Partitioning and Sparse Matrix
Ordering.
Homepage
of metis.
Comments:
-
Details in the manual
-
include file: parmetis.h
-
In linking add: -lmetis -lparmetis
BLAS
-
The BLAS library for single and dual proccesor Pentium Pro machines.
Comments:
-
Libraries for PII are under development.
-
View a quick reference manual
-
In linking add : -lblas or -lblas_single or -lblas_dual.
Currently -lblas links blas_single.
PBLAS
-
Parallel version of Blas.
Comments:
Lapack, Lapack90
-
Lapack routines and Lapack90 interface to them.
Comments:
-
Modules with interfaces are in /usr/local/include/lapack90_mod
-
View a quick reference manual
-
to view the man page : man lapack.
for a specific routine: man 'routine-name'
-
In linking add : [/usr/local/lib/libllapack90] /usr/local/lib/liblapack.a
/usr/local/lib/libblas_single.a
or : [/usr/local/lib/libllapack90] /usr/local/lib/liblapack.a /usr/local/lib/libblas_dual.a
/usr/lib/lpthread.a
ScaLapack
-
The parallel version of lapack seems to be installed.
Comments:
BlockSolve95 v3.0
-
The BlockSolve95 package contains routines for solving large sparse symmetric
systems of linear equations on massively parallel distributed memory systems
and networks of workstations.
Comments:
-
Details in the manual
-
to view the man pages: man 'routine-name'
-
include files in : /usr/local/include/BlockSolve95/
-
In linking add: -lBS95
-
examples available on request.
NAG f95 compiler and debugger (dbx90)
-
The NAGWare f95 Compiler is a full Fortran 95 compiler for a range of Unix
platforms. Derived from the NAGWare f90 Compiler, the world's first Fortran
90 compiler, f95 provides robust, well tested, widely ported compiler capabilities
for the Fortran 95 language. The f95 Compiler accepts fixed or free format
Fortran 95 input and a number of common Fortran 77 extensions. HPF code
is also compiled and checked though only single processor output is generated.
Comments:
-
to view the man pages: man nagf95, man nagdbx90
CVODE
-
(stiff or not) ODE integrator.
Comments:
-
written in C but callable from C or Fortran.
-
Details in the manual
-
include files in : /usr/local/include/cvode
-
In linking add: -lcvode
FFTW v2.0
-
THE FFT package (reference for all others). Uniprocessor version. MPI version
has not been tried, yet. Written in C, can be called from Fortran as every
C routine.
Homepage of FFTW.
Comments:
GNU Plotting utilities
-
Release 2.1.5 of the GNU plotutils (plotting utilities) package, including
release 1.5 of GNU libplot: a function library for two-dimensional device-independent
vector graphics, including vector graphics animations under the X Window
System.
Comments:
PETSc
-
PETSc the Portable Extensible Toolkit for Scientific computation is a suite
of uni- and parallel processor codes that are intended for the solution
of large scale problems modeled by partial differential equations. PETSc
employs the MPI standard for all message passing communication. The code
is written in a data-structure-neutral manner to enable easy reuse and
flexibility.
PETSc is easy to use for beginners. Moreover, its careful design allows
advanced users to have detailed control over the solution process. PETSc
includes an expanding suite of parallel linear and nonlinear equation solvers
that are easily used in application codes written in C, C++, and Fortran.
PETSc provides many of the mechanisms needed within parallel application
codes, such as simple parallel matrix and vector assembly routines that
allow the overlap of communication and computation. In addition, PETSc
includes growing support for distributed arrays. PETSc integrates a hierarchy
of components, thus enabling the user to employ the level of abstraction
that is most natural for a particular problem. Some of the components are
-
Mat -
-
a suite of data structures and code for the manipulation of parallel sparse
matrices,
-
PC -
-
a collection of preconditioners,
-
KSP -
-
data-structure-neutral implementations of many popular Krylov subspace
iterative methods,
-
SLES -
-
a higher-level interface for the solution of large-scale linear systems
and,
-
SNES -
-
data-structure-neutral implementations of Newton-like methods for nonlinear
systems,
-
TS -
-
timestepping code for scalably solving nonlinear ODEs arising from applying
general "methods of lines", techniques to time-dependent PDEs. Includes
support for pseudo-timestepping.
Manual: PETSc.ps.gz
Various Tutorials
Numerical Recipes (the text).
-
The full text of Numerical Recipes is available is ps files:
-
in C : Numerical_Recipes/C
-
in Fortran 77 : Numerical_Recipes/Fortran
EBOOK (in EBOOK).
-
contents of an electronic book for numerical algorithms.
(from the Computational Science Education Project)
F90
-
Collected tutorials, courses, FAQ, etc. here
-
Tutorial from the university of new mexico.
-
see also EBOOK/f90.ps.gz
HPF
-
Manual for version 2.0.
-
Collected tutorials, courses, FAQ, etc. here
G77.
-
Manual of the g77 compiler is g77.ps.gz.
MPI (in MPI).
-
Various documents about MPI. See MPI/README
Various Utilities
f2f90
-
A program to convert FORTRAN 77 fixed source form to FORTRAN 90 free source
form. Consult the man page for more information. Does not accept command
line arguments; use the program prompt to pass them.