Instructions for the Installation of HDF5 Software ================================================== This file provides instructions for installing the HDF5 software. If you have any problems with the installation, please see The HDF Group's support page at the following location: http://www.hdfgroup.org/services/support.html CONTENTS -------- 1. Obtaining HDF5 2. Quick installation 2.1. Windows 2.2. RedStorm (Cray XT3) 3. HDF5 dependencies 3.1. Zlib 3.2 Szip (optional) 3.3. MPI and MPI-IO 4. Full installation instructions for source distributions 4.1. Unpacking the distribution 4.1.1. Non-compressed tar archive (*.tar) 4.1.2. Compressed tar archive (*.tar.Z) 4.1.3. Gzip'd tar archive (*.tar.gz) 4.1.4. Bzip'd tar archive (*.tar.bz2) 4.2. Source versus build directories 4.3. Configuring 4.3.1. Specifying the installation directories 4.3.2. Using an alternate C compiler 4.3.3. Configuring for 64-bit support 4.3.4. Additional compilation flags 4.3.5. Compiling HDF5 wrapper libraries 4.3.6. Specifying other programs 4.3.7. Specifying other libraries and headers 4.3.8. Static versus shared linking 4.3.9. Optimization versus symbolic debugging 4.3.10. Parallel versus serial library 4.3.11. Threadsafe capability 4.3.12. Backward compatibility 4.4. Building 4.5. Testing 4.6. Installing HDF5 5. Using the Library 6. Support A. Warnings about compilers A.1. GNU (Intel platforms) A.2. DEC A.3. SGI (Irix64 6.2) A.4. Windows/NT B. Large (>2GB) versus small (<2GB) file capability C. Building and testing with other compilers C.1. Building and testing with Intel compilers C.2. Building and testing with PGI compilers ***************************************************************************** 1. Obtaining HDF5 The latest supported public release of HDF5 is available from ftp://ftp.hdfgroup.org/HDF5/current/src. For Unix and UNIX-like platforms, it is available in tar format compressed with gzip. For Microsoft Windows, it is in ZIP format. The HDF team also makes snapshots of the source code available on a regular basis. These snapshots are unsupported (that is, the HDF team will not release a bug-fix on a particular snapshot; rather any bug fixes will be rolled into the next snapshot). Furthermore, the snapshots have only been tested on a few machines and may not test correctly for parallel applications. Snapshots, in a limited number of formats, can be found on THG's development FTP server: ftp://ftp.hdfgroup.uiuc.edu/pub/outgoing/hdf5/snapshots 2. Quick installation For those who don't like to read ;-) the following steps can be used to configure, build, test, and install the HDF5 Library, header files, and support programs. For example, to install HDF5 version X.Y.Z at location /usr/local/hdf5, use the following steps. $ gunzip < hdf5-X.Y.Z.tar.gz | tar xf - $ cd hdf5-X.Y.Z $ ./configure --prefix=/usr/local/hdf5 $ make $ make check # run test suite. $ make install $ make check-install # verify installation. Some versions of the tar command support the -z option. In such cases, the first step above can be simplified to the following: $ tar zxf hdf5-X.Y.Z.tar.gz above refers to the configure flags appropriate to your installation. For example, to install HDF5 with the Fortran and C++ interfaces and with SZIP compression, the configure line might read as follows: $ ./configure --prefix=/usr/local/hdf5 --enable-fortran \ --enable-cxx --with-szlib=PATH_TO_SZIP In this case, PATH_TO_SZIP would be replaced with the path to the installed location of the SZIP library. 2.1. Windows Users of Microsoft Windows should see the INSTALL_Windows files for detailed instructions. 2.2. RedStorm (Cray XT3) Users of the Red Storm machine, after reading this file, should read the Red Storm section in the INSTALL_parallel file for specific instructions for the Red Storm machine. The same instructions would probably work for other Cray XT3 systems, but they have not been verified. 3. HDF5 dependencies 3.1. Zlib The HDF5 Library includes a predefined compression filter that uses the "deflate" method for chunked datasets. If zlib-1.1.2 or later is found, HDF5 will use it. Otherwise, HDF5's predefined compression method will degenerate to a no-op; the compression filter will succeed but the data will not be compressed. 3.2. Szip (optional) The HDF5 Library includes a predefined compression filter that uses the extended-Rice lossless compression algorithm for chunked datasets. For more information about Szip compression and license terms, see http://hdfgroup.org/doc_resource/SZIP/. The Szip source code can be obtained from the HDF5 Download page http://www.hdfgroup.org/HDF5/release/obtain5.html#extlibs. Building instructions are available with the Szip source code. The HDF Group does not distribute separate Szip precompiled libraries, but the HDF5 binaries available from http://www.hdfgroup.org/HDF5/release/obtain5.html include the Szip encoder enabled binary for the corresponding platform. To configure the HDF5 Library with the Szip compression filter, use the '--with-szlib=/PATH_TO_SZIP' flag. For more information, see section 4.3.7, "Specifying other libraries and headers." Please notice that if HDF5 configure cannot find a valid Szip library, configure will not fail; in this case, the compression filter will not be available to the applications. To check if Szip compression was successfully configured in, check the "I/O filters (external):" line in the configure output, summary section, printed to the standard output. 3.3. MPI and MPI-IO The parallel version of the library is built upon the foundation provided by MPI and MPI-IO. If these libraries are not available when HDF5 is configured, only a serial version of HDF5 can be built. 4. Full installation instructions for source distributions 4.1. Unpacking the distribution The HDF5 source code is distributed in a variety of formats which can be unpacked with the following commands, each of which creates an 'hdf5-X.Y.Z' directory, where X.Y.Z is the HDF5 version numbers. 4.1.1. Non-compressed tar archive (*.tar) $ tar xf hdf5-X.Y.Z.tar 4.1.2. Compressed tar archive (*.tar.Z) $ uncompress -c < hdf5-X.Y.Z.tar.Z | tar xf - Or $ tar Zxf hdf5-X.Y.Z.tar.Z 4.1.3. Gzip'd tar archive (*.tar.gz) $ gunzip < hdf5-X.Y.Z.tar.gz | tar xf - Or $ tar zxf hdf5-X.Y.Z.tar.gz 4.1.4. Bzip'd tar archive (*.tar.bz2) $ bunzip2 < hdf5-X.Y.Z.tar.bz2 | tar xf - Or $ tar jxf hdf5-X.Y.Z.tar.bz2 4.2. Source versus build directories On most systems the build can occur in a directory other than the source directory, allowing multiple concurrent builds and/or read-only source code. In order to accomplish this, one should create a build directory, cd into that directory, and run the `configure' script found in the source directory (configure details are below). For example, $ mkdir built-fortran $ cd build-fortran $ ../hdf5-X.Y.Z/configure --enable-fortran ... Unfortunately, this does not work on recent Irix platforms (6.5? and later) because that `make' does not understand the VPATH variable. However, HDF5 also supports Irix `pmake' which has a .PATH target which serves a similar purpose. Here's what the Irix man pages say about VPATH, the facility used by HDF5 makefiles for this feature: The VPATH facility is a derivation of the undocumented VPATH feature in the System V Release 3 version of make. System V Release 4 has a new VPATH implementation, much like the pmake(1) .PATH feature. This new feature is also undocumented in the standard System V Release 4 manual pages. For this reason it is not available in the IRIX version of make. The VPATH facility should not be used with the new parallel make option. 4.3. Configuring HDF5 uses the GNU autoconf system for configuration, which detects various features of the host system and creates the Makefiles. On most systems it should be sufficient to say: $ ./configure Or $ sh configure The configuration process can be controlled through environment variables, command-line switches, and host configuration files. For a complete list of switches type: $ ./configure --help The host configuration files are located in the `config' directory and are based on architecture name, vendor name, and/or operating system which are displayed near the beginning of the `configure' output. The host config file influences the behavior of configure by setting or augmenting shell variables. 4.3.1. Specifying the installation directories The default installation location is the HDF5 directory created in the build directory. Typing `make install' will install the HDF5 Library, header files, examples, and support programs in hdf5/lib, hdf5/include, hdf5/doc/hdf5/examples, and hdf5/bin. To use a path other than hdf5, specify the path with the `--prefix=PATH' switch: $ ./configure --prefix=/usr/local If shared libraries are being built (the default), the final home of the shared library must be specified with this switch before the library and executables are built. HDF5 can be installed into a different location than the prefix specified at configure time; see section 4.6, "Installing HDF5," for more details. 4.3.2. Using an alternate C compiler By default, configure will look for the C compiler by trying `gcc' and `cc'. However, if the environment variable "CC" is set then its value is used as the C compiler. For instance, one would use the following line to specify the native C compiler on a system that also has the GNU gcc compiler (users of csh and derivatives will need to prefix the commands below with `env'): $ CC=cc ./configure A parallel version of HDF5 can be built by specifying `mpicc' as the C compiler. (The `--enable-parallel' flag documented below is optional in this case.) Using the `mpicc' compiler will insure that the correct MPI and MPI-IO header files and libraries are used. $ CC=/usr/local/mpi/bin/mpicc ./configure 4.3.3. Configuring for 64-bit support Several machine architectures support 32-bit or 64-bit binaries. The options below describe how to enable support for different options. On Irix64, the default compiler is `cc'. To use an alternate compiler, specify it with the CC variable: $ CC='cc -n32' ./configure Similarly, users compiling on a Solaris machine and desiring to build the distribution with 64-bit support should specify the correct flags with the CC variable: $ CC='cc -m64' ./configure To configure AIX 64-bit support including the Fortran and C++ APIs, (Note: need to set $AR to 'ar -X 64'.) Serial: $ CFLAGS=-q64 FCFLAGS=-q64 CXXFLAGS=-q64 AR='ar -X 64'\ ./configure --enable-fortran Parallel: (C++ not supported with parallel) $ CFLAGS=-q64 FCFLAGS=-q64 AR='ar -X 64'\ ./configure --enable-fortran 4.3.4. Additional compilation flags If addtional flags must be passed to the compilation commands, specify those flags with the CFLAGS variable. For instance, to enable symbolic debugging of a production version of HDF5, one might say: $ CFLAGS=-g ./configure --enable-build-mode=production 4.3.5. Compiling HDF5 wrapper libraries One can optionally build the Fortran and/or C++ interfaces to the HDF5 C library. By default, both options are disabled. To build them, specify `--enable-fortran' and `--enable-cxx', respectively. $ ./configure --enable-fortran $ ./configure --enable-cxx Configuration will halt if a working Fortran 90 or 95 compiler or C++ compiler is not found. Currently, the Fortran configure tests for these compilers in order: f90, pgf90, f95. To use an alternate compiler specify it with the FC variable: $ FC=/usr/local/bin/g95 ./configure --enable-fortran Note: The Fortran and C++ interfaces are not supported on all the platforms the main HDF5 Library supports. Also, the Fortran interface supports parallel HDF5 while the C++ interface does not. Note: See sections 4.7 and 4.8 for building the Fortran library with Intel or PGI compilers. 4.3.6. Specifying other programs The build system has been tuned for use with GNU make but also works with other versions of make. If the `make' command runs a non-GNU version but a GNU version is available under a different name (perhaps `gmake'), then HDF5 can be configured to use it by setting the MAKE variable. Note that whatever value is used for MAKE must also be used as the make command when building the library: $ MAKE=gmake ./configure $ gmake The `AR' and `RANLIB' variables can also be set to the names of the `ar' and `ranlib' (or `:') commands to override values detected by configure. The HDF5 Library, include files, and utilities are installed during `make install' (described below) with a BSD-compatible install program detected automatically by configure. If none is found, the shell script bin/install-sh is used. Configure does not check that the install script actually works; if a bad install is detected on your system (e.g., on the ASCI blue machine as of March 2, 1999) you have two choices: 1. Copy the bin/install-sh program to your $HOME/bin directory, name it `install', and make sure that $HOME/bin is searched before the system bin directories. 2. Specify the full path name of the `install-sh' program as the value of the INSTALL environment variable. Note: do not use `cp' or some other program in place of install because the HDF5 makefiles also use the install program to change file ownership and/or access permissions. 4.3.7. Specifying other libraries and headers Configure searches the standard places (those places known by the systems compiler) for include files and header files. However, additional directories can be specified by using the CPPFLAGS and/or LDFLAGS variables: $ CPPFLAGS=-I/home/robb/include \ LDFLAGS=-L/home/robb/lib \ ./configure HDF5 uses the zlib library to support the HDF5 deflate data compression filter. Configure searches the standard places (plus those specified above with the CPPFLAGS and LDFLAGS variables) for the zlib headers and library. The search can be disabled by specifying `--without-zlib' or alternate directories can be specified with `--with-zlib=INCDIR,LIBDIR' or through the CPPFLAGS and LDFLAGS variables: $ ./configure --with-zlib=/usr/unsup/include,/usr/unsup/lib $ CPPFLAGS=-I/usr/unsup/include \ LDFLAGS=-L/usr/unsup/lib \ ./configure HDF5 includes Szip as a predefined compression method (see 3.2). To enable Szip compression, the HDF5 Library must be configured and built using the Szip Library: $ ./configure --with-szlib=/Szip_Install_Directory 4.3.8. Static versus shared linking The build process will create static libraries on all systems and shared libraries on systems that support dynamic linking to a sufficient degree. Either form of the library may be suppressed by saying `--disable-static' or `--disable-shared'. $ ./configure --disable-shared Shared C++ and Fortran libraries will be built if shared libraries are enabled. To build only statically linked executables on platforms which support shared libraries, use the `--enable-static-exec' flag. $ ./configure --enable-static-exec 4.3.9. Optimization versus symbolic debugging The library can be compiled to provide symbolic debugging support so it can be debugged with gdb, dbx, ddd, etc., or it can be compiled with various optimizations. To compile for symbolic debugging (the default for snapshots), say `--enable-build-mode=production'; to compile with optimizations (the default for supported public releases), say `--enable-build-mode=production'. On some systems the library can also be compiled for profiling with gprof by saying `--enable-production=profile'. $ ./configure --enable-build-mode=debug #symbolic debugging $ ./configure --enable-build-mode=production #optimized code $ ./configure --enable-production=profile #for use with gprof Regardless of whether support for symbolic debugging is enabled, the library can also perform runtime debugging of certain packages (such as type conversion execution times and extensive invariant condition checking). To enable this debugging, supply a comma-separated list of package names to the `--enable-internal-debug' switch. See "Debugging HDF5 Applications" for a list of package names: http://www.hdfgroup.org/HDF5/doc/H5.user/Debugging.html Debugging can be disabled by saying `--disable-internal-debug'. The default debugging level for snapshots is a subset of the available packages; the default for supported releases is no debugging (debugging can incur a significant runtime penalty). $ ./configure --enable-internal-debug=s,t #debug only H5S and H5T $ ./configure --enable-internal-debug #debug normal packages $ ./configure --enable-internal-debug=all #debug all packages $ ./configure --disable-internal-debug #no debugging HDF5 can also print a trace of all API function calls, their arguments, and the return values. To enable or disable the ability to trace the API say `--enable-trace' (the default for snapthots) or `--disable-trace' (the default for public releases). The tracing must also be enabled at runtime to see any output (see "Debugging HDF5 Applications," reference above). 4.3.10. Parallel versus serial library The HDF5 Library can be configured to use MPI and MPI-IO for parallelism on a distributed multi-processor system. Read the file INSTALL_parallel for detailed explanations. 4.3.11. Threadsafe capability The HDF5 Library can be configured to be thread-safe (on a very large scale) with the `--enable-threadsafe' flag to the configure script. Some platforms may also require the '-with-pthread=INC,LIB' (or '--with-pthread=DIR') flag to the configure script. For further details, see "HDF5 Thread Safe Library": http://www.hdfgroup.org/HDF5/doc/TechNotes/ThreadSafeLibrary.html 4.3.12. Backward compatibility The 1.10 version of the HDF5 Library can be configured to operate identically to the v1.8 library with the --with-default-api-version=v18 configure flag, or identically to the v1.6 library with the --with-default-api-version=v16 configure flag. This allows existing code to be compiled with the v1.10 library without requiring immediate changes to the application source code. For addtional configuration options and other details, see "API Compatibility Macros in HDF5": https://support.hdfgroup.org/HDF5/doc/RM/APICompatMacros.html 4.4. Building The library, confidence tests, and programs can be built by saying just: $ make Note that if you have supplied some other make command via the MAKE variable during the configuration step, that same command must be used here. When using GNU make, you can add `-j -l6' to the make command to compile in parallel on SMP machines. Do not give a number after the `-j' since GNU make will turn it off for recursive invocations of make. $ make -j -l6 4.5. Testing HDF5 comes with various test suites, all of which can be run by saying $ make check To run only the tests for the library, change to the `test' directory before issuing the command. Similarly, tests for the parallel aspects of the library are in `testpar' and tests for the support programs are in `tools'. The `check' consists of two sub-tests, check-s and check-p, which are for serial and parallel tests, respectively. Since serial tests and parallel tests must be run with single and multiple processes respectively, the two sub-tests work nicely for batch systems in which the number of processes is fixed per batch job. One may submit one batch job, requesting 1 process, to run all the serial tests by "make check-s"; and submit another batch job, requesting multiple processes, to run all the parallel tests by "make check-p". Temporary files will be deleted by each test when it completes, but may continue to exist in an incomplete state if the test fails. To prevent deletion of the files, define the HDF5_NOCLEANUP environment variable. The HDF5 tests can take a long time to run on some systems. To perform a faster (but less thorough) test, set the HDF5TestExpress environment variable to 2 or 3 (with 3 being the shortest run). To perform a longer test, set HDF5TestExpress to 0. 1 is the default. 4.6. Installing HDF5 The HDF5 Library, include files, and support programs can be installed in a (semi-)public place by saying `make install'. The files are installed under the directory specified with `--prefix=DIR' (default is 'hdf5') in directories named `lib', `include', and `bin'. The directories, if not existing, will be created automatically, provided the mkdir command supports the -p option. If `make install' fails because the install command at your site somehow fails, you may use the install-sh that comes with the source. You will need to run ./configure again. $ INSTALL="$PWD/bin/install-sh -c" ./configure ... $ make install If you want to install HDF5 in a location other than the location specified by the `--prefix=DIR' flag during configuration (or instead of the default location, `hdf5'), you can do that by running the deploy script: $ bin/deploy NEW_DIR This will install HDF5 in NEW_DIR. Alternately, you can do this manually by issuing the command: $ make install prefix=NEW_DIR where NEW_DIR is the new directory where you wish to install HDF5. If you do not use the deploy script, you should run h5redeploy in NEW_DIR/bin directory. This utility will fix the h5cc, h5fc and h5c++ scripts to reflect the new NEW_DIR location. The library can be used without installing it by pointing the compiler at the `src' and 'src/.libs' directory for include files and libraries. However, the minimum which must be installed to make the library publicly available is: The library: ./src/.libs/libhdf5.a The public header files: ./src/H5*public.h, ./src/H5public.h ./src/H5FD*.h except ./src/H5FDprivate.h, ./src/H5api_adpt.h The main header file: ./src/hdf5.h The configuration information: ./src/H5pubconf.h The support programs that are useful are: ./tools/h5ls/h5ls (list file contents) ./tools/h5dump/h5dump (dump file contents) ./tools/misc/h5repart (repartition file families) ./tools/misc/h5debug (low-level file debugging) ./tools/h5import/h5import (imports data to HDF5 file) ./tools/h5diff/h5diff (compares two HDF5 files) ./tools/gifconv/h52gif (HDF5 to GIF converter) ./tools/gifconv/gif2h5 (GIF to HDF5 converter) 5. Using the Library Please see the "HDF5 User's Guide" and the "HDF5 Reference Manual": http://www.hdfgroup.org/HDF5/doc/ Most programs will include and link with -lhdf5. Additional libraries may also be necessary depending on whether support for compression, etc., was compiled into the HDF5 Library. A summary of the HDF5 installation can be found in the libhdf5.settings file in the same directory as the static and/or shared HDF5 Libraries. 6. Support Support is described in the README file. ***************************************************************************** APPENDIX ***************************************************************************** A. Warnings about compilers Output from the following compilers should be extremely suspected when used to compile the HDF5 Library, especially if optimizations are enabled. In all cases, HDF5 attempts to work around the compiler bugs. A.1. GNU (Intel platforms) Versions before 2.8.1 have serious problems allocating registers when functions contain operations on `long long' datatypes. A.2. COMPAQ/DEC The V5.2-038 compiler (and possibly others) occasionally generates incorrect code for memcpy() calls when optimizations are enabled, resulting in unaligned access faults. HDF5 works around the problem by casting the second argument to `char *'. The Fortran module (5.4.1a) fails in compiling some Fortran programs. Use 5.5.0 or higher. A.3. SGI (Irix64 6.2) The Mongoose 7.00 compiler has serious optimization bugs and should be upgraded to MIPSpro 7.2.1.2m. Patches are available from SGI. A.4. Windows/NT The Microsoft Win32 5.0 compiler is unable to cast unsigned long long values to doubles. HDF5 works around this bug by first casting to signed long long and then to double. A link warning: defaultlib "LIBC" conflicts with use of other libs appears for debug version of VC++ 6.0. This warning will not affect building and testing HDF5 Libraries. B. Large (>2GB) versus small (<2GB) file capability In order to read or write files that could potentially be larger than 2GB, it is necessary to use the non-ANSI `long long' data type on some platforms. However, some compilers (e.g., GNU gcc versions before 2.8.1 on Intel platforms) are unable to produce correct machine code for this datatype. C. Building and testing with other compilers C.1. Building and testing with Intel compilers When Intel compilers are used (icc or ecc), you will need to modify the generated "libtool" program after configuration is finished. On or around line 104 of the libtool file, there are lines which look like: # How to pass a linker flag through the compiler. wl="" Change these lines to this: # How to pass a linker flag through the compiler. wl="-Wl," UPDATE: This is now done automatically by the configure script. However, if you still experience a problem, you may want to check this line in the libtool file and make sure that it has the correct value. * To build the Fortran library using Intel compiler on Linux 2.4, one has to perform the following steps: x Use the -fpp -DDEC$=DEC_ -DMS$=MS_ compiler flags to disable DEC and MS compiler directives in source files in the fortran/src, fortran/test, and fortran/examples directories. E.g., setenv F9X 'ifc -fpp -DDEC$=DEC_ -DMS$=MS_' Do not use double quotes since $ is interpreted in them. x If Version 6.0 of Fortran compiler is used, the build fails in the fortran/test directory and then in the fortran/examples directory. To proceed, edit the work.pcl files in those directories to contain two lines: work.pc ../src/work.pc x Do the same in the fortran/examples directory. x A problem with work.pc files was resolved for the newest version of the compiler (7.0). * To build the Fortran library on IA32, follow the steps described above, except that the DEC and MS compiler directives should be removed manually or use a patch from HDF FTP server: ftp://ftp.hdfgroup.org/HDF5/current/ C.2. Building and testing with PGI compilers When PGI C and C++ compilers are used (pgcc or pgCC), you will need to modify the generated "libtool" program after configuration is finished. On or around line 104 of the libtool file, there are lines which look like this: # How to pass a linker flag through the compiler. wl="" Change these lines to this: # How to pass a linker flag through the compiler. wl="-Wl," UPDATE: This is now done automatically by the configure script. However, if you still experience a problem, you may want to check this line in the libtool file and make sure that it has the correct value. To build the HDF5 C++ Library with pgCC (version 4.0 and later), set the environment variable CXX to "pgCC -tlocal" setenv CXX "pgCC -tlocal" before running the configure script.