Version 105 (modified by knoop, 13 months ago) (diff) |
---|
Download & Installation
Download
If you are interested in using or just testing the code, you can download all available PALM releases from our GitLab Server
Software requirements
In order to successfully install PALM, please meet the following software requirements:
- A Linux Operating System (OS). We test PALM on Debian-based OS only and can not help if you use a different OS.
- The bash shell (available at /bin/bash).
- A recent Fortran and C++ Compiler (GNU, Intel, Cray, PGI, NEC).
- The build automation tools cmake and make (for library detection and build coordination).
- The Message Passing Interface (MPI) library with MPI-3 support (compiled with the same compiler as PALM).
- A NetCDF library not earlier than 3.6.3 (compiled with the same Compiler as PALM).
- The FFT library FFTW (PALM also comes with a build-in FFT but with less performance).
- Python 3 (needed by the GUI and other helper routines).
- PyQt5 (needed by the GUI).
- The graphic-package NCL from NCAR (needed by the data visualization tool palmplot).
- The FLEX library BISON parser generator (needed by the chemistry tool kpp4palm)
The README.md file that ships with the PALM Model System release Package provides a one-liner that triggers the installation of all requirements on most Debian-based Linux Distributions like Ubuntu etc.
Very important: It is essential that your NetCDF and MPI library has been built with the same Fortran compiler that is used to compile PALM. Furthermore, in case of a NetCDF4 library with parallel I/O support, the NetCDF library needs to be build with the same MPI library as used for compiling PALM.
PALM installation
From our GitLab Server the PALM Model System can be downloaded as a zip or tar archive. Please unpack the content to a directory of your choice and open a terminal inside that directory. Please follow the installation instructions described in the README.md file that ships with the PALM Model System release Package. The installation instructions mention, that an <install-prefix> needs to be chosen. Please be aware that this directory will become the storage location for all your PALM input and output data. So you might wand to make sure to have sufficient disk space. In order to create a directory structure similar to the one that was standard with the old SVN based repository, please unpack the zip or tar archive to the directory ~/palm/current_version/palm_model_system and choose ~/palm/current_version as the <install-prefix>. In any case, please make sure no errors have occurred by carefully checking the output of the installation script.
After the installation is finished you should have the following files/folders inside the installation directory ~/palm/current_version:
- The PALM Model System code repository palm_model_system/.
- A configuration file .palm.config.default.
- A folder named bin/ that contains all the executables.
- A folder named rrtmg/ that contains the RRTMG library.
- A folder named MAKE_DEPOSITORY_default/ that contains the pre-compiled PALM code.
For a possible fast track to execute your first simulation, execute these commands:
export PATH=~/palm/current_version/bin:${PATH} cd ~/palm/current_version mkdir -p JOBS/example_cbl/INPUT cp palm_model_system/packages/palm/model/tests/cases/example_cbl/INPUT/example_cbl_p3d JOBS/example_cbl/INPUT/ palmrun -r example_cbl -c default -a "d3#" -X 4 -v -z
If you encounter any errors during the installation, try following the full manual installation
Note: With the default install script, you can just run PALM interactively on the local computer. It do not allow to setup the creation of batch jobs and restart runs on any local or remote computers. Therefore, theis method is only suitable / recommended for users who like to do a quick test of PALM. If you like to work with PALM on a HPC cluster with batch scheduling, please also follow the instructions for the full manual installation
Other installation methods
The following additional methods to install and run PALM on Linux/UNIX systems are available:
By the way, you can / should use the default install script as a basis for a complete manual installation.
The manual installation gives full access to all PALM features. It should be chosen to setup PALMs ability to run in batch mode on a local or remote computer (see software requirements). It allows to submit a job on the local host while instructing palmrun to execute the job on your desired remote host. The PALM output is then automatically transferred back to the local computer. The manual installation can of course also be used to work with PALM in interactive mode on the local computer. The installation process for the manual installation may require a valid account on the remote host as well.
The examples given in this chapter refer to a local installation of PALM on an IMUK Linux workstation (local host) and for running batch jobs on the Cray XC40 system of HLRN (remote host).
In case of problems with the installation, please check the FAQ Installation Questions first.
Supported Hardware
PALM runs on all modern multicore workstation processors and on many different HPC systems. Furthermore, PALM has recently been ported to Graphics Processing Units (GPU) like the NVidia Tesla series and to manycore processor architectures like the Intel Xeon Phi Knights Landing. Please see details
Regarding batch support, PALM can be configured to be used on arbitrary supercomputers (see full manual installation).
Attachments (1)
- palm_installer (53.8 KB) - added by knoop 8 years ago.
Download all attachments as: .zip