moved install instructions to file INSTALL

This commit is contained in:
2009-01-15 15:18:37 +00:00
parent 11bc7db824
commit 6633dace44
+56 -417
View File
@@ -1,418 +1,57 @@
==============================================================================
README file for H5Part configure
==============================================================================
+ 0. HDF5 library
-----------------
Make sure you have a working version of the HDF5 library in your computer. If you plan
to use parallel I/O you need to use a parallel HDF5 version. If you don't have the
library download the sources from http://hdf.ncsa.uiuc.edu/HDF5/
+ 1. Quick start (If you feel lucky...)
----------------------------------------
For many platforms, where compilers are installed in the "default" location
and all the environment variables are set correctly, it should be sufficient
to type:
./configure [OPTIONS] && make [install]
to have the libraries and test program compiled.
==============================================================================
+ 2. configure line options
----------------------------
(1) Enable options
--enable-fortran
--enable-parallel
--enable-tools
--enable-python
--enable-64 (only for AIX and Irix)
(2) Setting compilers related variables manually
To assign environment variables (e.g., CC, CFLAGS...), specify them as
VAR=VALUE. See below for descriptions of some of the useful variables.
Compiler environment variables:
CC C compiler command
CXX C++ compiler command
FC Fortran compiler command
(3) Setting prefix for installation manually
--prefix=PREFIX install files in PREFIX
For more information, type:
./configure --help
==============================================================================
+ 3. Platform specific settings for compiling libraries & test programs
------------------------------------------------------------------------
Building
make : builds the libraries and test programs
make install: builds and installs
make clean : removes extraneous object files
make distclean : returns configuration to unconfigured state
As tested on hosts:
(1) Davinci <davinci.nersc.gov> (Linux"SUSE"-ia64)
---------------------------------------------------
FIRST THING FIRST!! Type:
module load intel
to have the proper compilers set up for use.
module load hdf5_par
[NOTE: module load hdf5 does not work at this moment.
Possible installation error of HDF5 on Davinci.]
Brief profile:
CC = icc
CXX = icc
FC = ifort
MPICC = icc
MPICXX = icc
MPIFC = ifort
If default doesn't work, try:
For C shell:
env CC=icc CXX=icc FC=ifort MPICC=icc MPICXX=icc MPIFC=ifort ./configure && make install
For Korn or Bourne shell:
CC=icc CXX=icc FC=ifort MPICC=icc MPICXX=icc MPIFC=ifort ./configure && make install
And, of course, you should add configure options, such as --enable-fortran --enable-parallel.
On this host, executing:
For C shell:
env CC=icc CXX=icc FC=ifort MPICC=icc MPICXX=icc MPIFC=ifort ./configure --enable-fortran --enable-parallel && make install
For Korn or Bourne shell:
CC=icc CXX=icc FC=ifort MPICC=icc MPICXX=icc MPIFC=ifort ./configure --enable-fortran --enable-parallel && make install
has functioned correctly.
To run test program, go to test directory:
(i) Serial test
run:
./H5testF
./H5PartTest
(ii) Parallel test
run:
mpirun 2 H5testFpar
mpirun 2 H5PartTestP
NOTE: Number 2 above is a arbitrary number that indicates the number of processes.
(2) Jacquard <jacquard.nersc.gov> (Linux"SUSE"-x86_64)
-------------------------------------------------------
Brief profile:
CC = pathcc
CXX = pathCC
FC = pathf90
MPICC = mpicc
MPICXX = mpicxx
MPIFC = mpif90
If default doesn't work, try:
For C shell:
env CC=pathcc CXX=pathCC FC=pathf90 MPICC=mpicc MPICXX=mpicxx MPIFC=mpif90 ./configure && make install
For Korn or Bourne shell:
CC=pathcc CXX=pathCC FC=pathf90 MPICC=mpicc MPICXX=mpicxx MPIFC=mpif90 ./configure && make install
And, of course, you should add configure options, such as --enable-fortran --enable-parallel.
On this host, executing:
For C shell:
env CC=pathcc CXX=pathCC FC=pathf90 MPICC=mpicc MPICXX=mpicxx MPIFC=mpif90 ./configure --enable-fortran --enable-parallel && make install
For Korn or Bourne shell:
CC=pathcc CXX=pathCC FC=pathf90 MPICC=mpicc MPICXX=mpicxx MPIFC=mpif90 ./configure --enable-fortran --enable-parallel && make install
has functioned correctly.
To run test program, go to test directory:
(i) Serial test
run:
./H5testF
./H5PartTest
(ii) Parallel test
In Jacquard, we need a script to run parallel programs, since Jacquard does not support interactive parallel processing. Instead, we need to submit the task to a queue. A sample script (run_H5testFpar.scr) may look like below:
+++++++++++++++++++++++++++++++++++++++
#!/bin/csh
#PBS -l nodes=1:ppn=2,walltime=00:05:00
#PBS -N H5testFpar
#PBS -o H5testFpar.out
#PBS -e H5testFpar.err
#PBS -q debug
#PBS -A mpccc
#PBS -V
setenv PBS_OWORKDIR /home/H5Part/test
cd $PBS_O_WORKDIR
mpirun -np 2 ./H5testFpar
+++++++++++++++++++++++++++++++++++++++
Then, we need to submit the task to a queue by running:
qsub run_H5testFpar.scr
After processed, the result will appear in H5testFpar.out, and error messages will appear in H5testFpar.err.
(3) Bassi <bassi.nersc.ogv> (AIX5.3.0.0-Power 3)
-------------------------------------------------
FIRST THING FIRST!! Type:
module load gcc
to have the proper compilers set up for use.
Brief profile:
CC = cc_r
CXX = cc_r
FC = xlf_r
MPICC = mpcc_r
MPICXX = mpcc_r
MPIFC = mpxlf_r
If default doesn't work, try:
For C shell:
env CC=cc_r CXX=cc_r FC=xlf_r MPICC=mpcc_r MPICXX=mpcc_r MPIFC=mpxlf_r ./configure && make install
For Korn or Bourne shell:
CC=cc_r CXX=cc_r FC=xlf_r MPICC=mpcc_r MPICXX=mpcc_r MPIFC=mpxlf_r ./configure && make install
And, of course, you should add configure options, such as --enable-fortran --enable-parallel.
On this host, executing:
For C shell:
env CC=cc_r CXX=cc_r FC=xlf_r MPICC=mpcc_r MPICXX=mpcc_r MPIFC=mpxlf_r ./configure --enable-fortran --enable-parallel && make install
For Korn or Bourne shell:
CC=cc_r CXX=cc_r FC=xlf_r MPICC=mpcc_r MPICXX=mpcc_r MPIFC=mpxlf_r ./configure --enable-fortran --enable-parallel && make install
has functioned correctly.
To run test program, go to test directory:
(i) Serial test
run:
./H5testF
./H5PartTest
(ii) Parallel test
run:
unsetenv MP_SINGLE_THREAD
poe ./H5PartTestP -procs 2 -nodes 1
poe ./H5testFpar -procs 2 -nodes 1
For more details on how to run parallel programs on Bassi, go to:
http://www.nersc.gov/nusers/resources/bassi/running_jobs/
(4) Starsky <starsky.lbl.gov> (Darwin8.5.0-PowerPC)
----------------------------------------------------
FIRST THING FIRST!! Set environment variables: (in tcsh)
setenv LD_LIBRARY_PATH /usr/local/g95/lib/gcc-lib/powerpc-apple-darwin8.5.0/4.0.3/
setenv PATH ${PATH}:/usr/local/g95/bin
NOTE: This environment variable setting is specifit to Starsky. It is to make configure find the fortran compiler.
Brief profile:
CC = gcc
CXX = g++
FC = g95
If default doesn't work, try:
For C shell:
env CC=gcc CXX=g++ FC=g95 ./configure && make install
For Korn or Bourne shell:
CC=gcc CXX=g++ FC=g95 ./configure && make install
And, of course, you should add configure options, such as --enable-fortran --enable-parallel.
On this host, executing:
For C shell:
env CC=gcc CXX=g++ FC=g95 ./configure --enable-fortran && make install
For Korn or Bourne shell:
CC=gcc CXX=g++ FC=g95 ./configure --enable-fortran && make install
has functioned correctly.
To run test program, go to test directory:
(i) Serial test
run:
./H5testF
./H5PartTest
(5) Linux AMD64 Visualization group workstations (Linux"SUSE"-x86_64)
---------------------------------------------------------
Brief profile:
CC = gcc
CXX = g++
FC = g95
If default doesn't work, try:
For C shell:
env CC=gcc CXX=g++ FC=g95 ./configure && make install
For Korn or Bourne shell:
CC=gcc CXX=g++ FC=g95 ./configure && make install
And, of course, you should add configure options, such as --enable-fortran --enable-parallel.
On this host, executing:
For C shell:
env CC=gcc CXX=g++ FC=g95 ./configure --enable-fortran && make install
For Korn or Bourne shell:
CC=gcc CXX=g++ FC=g95 ./configure --enable-fortran && make install
has functioned correctly.
To run test program, go to test directory:
(i) Serial test
run:
./H5testF
./H5PartTest
(6) Cray XT3 @ CSCS gele
1) modules/3.1.6 9) PrgEnv-pgi/1.4.26 17) xt-lustre-ss/1.4.26
2) MySQL/4.0.26 10) xt-pbs/5.3.4 18) Base-opts/1.4.26
3) acml/3.0 11) xt-service/1.4.26 19) subversion/1.3.2
4) pgi/6.1.4 12) xt-libc/1.4.26 20) zlib/1.2.3
5) totalview/7.2.0 13) xt-os/1.4.26 21) szip/2.0
6) xt-libsci/1.4.26 14) xt-catamount/1.4.26 22) hdf5/1.6.5
7) xt-mpt/1.4.26 15) xt-boot/1.4.26
8) xt-pe/1.4.26 16) xt-crms/1.4.26
CFLAGS="-I$H5HOME/include -DF77_SINGLE_UNDERSCORE" LDFLAGS="-L$H5HOME/lib -L$ZHOME/lib" ./configure --enable-parallel --enable-fortran
(7) merlin00 / merlin3
Currently Loaded Modulefiles:
1) hdf5/hdf5-1.6.5 2) root/root-5.10.00 3) mpi/mpich2-1.0.3-pgi-6.1 4) pgi/pgi_64-6.1
FC=mpif90 ./configure --enable-parallel --enable-fortran
==============================================================================
+ 4. Trouble shooting (Things to check for...)
--------------------------------------------
(0) Have you set the LD_LIBRARY_PATH?
---Some systems require the user to manually set the environment variable, LD_LIBRARY_PATH. To do so:
set the environment variable HDF5ROOT or PHDF5ROOT to point to your installation of HDF5 (serial and/or parallel respectively).
For Korn or Bourne shell:
LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:${HDF5ROOT}/lib;export LD_LIBRARY_PATH
LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:{PHDF5ROOT}/lib;export LD_LIBRARY_PATH
For C shell:
setenv LD_LIBRARY_PATH ${LD_LIBRARY_PATH}:${HDF5ROOT}/lib
setenv LD_LIBRARY_PATH ${LD_LIBRARY_PATH}:${PHDF5ROOT}/lib
Note that the values of HDF5ROOT & PHDF5ROOT will be in the "summary" section when you run configure.
Not properly set LD_LIBRARY_PATH value may result in a runtime error:
./H5PartTest: error while loading shared libraries: libhdf5.so.0: cannot open shared object file: No such file or directory
(1) Are the compilers set correctly?
---Check with: which COMPILER
(2) Have you installed HDF5?
---You can get it at: http://vis.lbl.gov/Research/AcceleratorSAPP/index.html
(3) Have you loaded the proper modules?
---It is necessary in Davinci & Bassi, and configure && make install may not work without it.
For more information on using the H5Part API, see the H5Part website:
http://www-vis.lbl.gov/Research/AcceleratorSAPP/
Motivation
----------
Particle based simulations of accelerator beam-lines, especially in six
dimensional phase space, generate vast amounts of data. Even though a subset
of statistical information regarding phase space or analysis needs to be
preserved, reading and writing such enormous restart files on massively
parallel supercomputing systems remains challenging.
H5Part is a very simple data storage schema and provides an API that
simplifies the reading/writing of the data to the HDF5 file format. An
important foundation for a stable visualization and data analysis environment
is a stable and portable file storage format and its associated APIs. The
presence of a "common file storage format," including associated APIs, will
help foster a fundamental level of interoperability across the project's
software infrastructure. It will also help ensure that key data analysis
capabilities are present during the earliest phases of the software development
effort.
H5Part is built on top of the HDF5 (Hierarchical Data Format) library. HDF5
offers a self-describing machine-independent binary file format that supports
scalable parallel I/O performance for MPI codes on a variety of supercomputing
systems, and works equally well on laptop computers. The API is available for
C, C++, and Fortran codes. The H5Part file format and APIs enable disparate
research groups with different simulation implementations to transparently
share datasets and data analysis tools. For instance, the common file format
will enable groups that depend on completely different simulation
implementations to share data analysis tools.
File Organization and API
-------------------------
The proposed file storage format uses HDF5 for the lowlevel file storage and a
simple API to provide a high-level interface to that file format. A programmer
can either use the H5Part API to access the data files or write directly to the
file format using some simple conventions for organizing and naming the objects
stored in the file.
In order to store Particle Data in the HDF5 file format, we have formalized the
hierarchical arrangement of the datasets and naming conventions for the groups
and associated datasets. The sample H5Part API formally encodes these
conventions in order to provide a simple and uniform way to access these files
from C, C++, and Fortran codes. The API makes it easier to write very portable
data adaptors for visualization tools in order to expand the number of tools
available to access the data. Even so, anyone can use the HDF5 h5ls utility to
examine the organization of the H5Part files and even write their own
HDF5-based interface for reading and writing the file format. The standards
offered by the sample API are completely independent of the standard for
organizing data within the file.The file format supports the storage of
multiple timesteps of datasets that contain multiple fields.