Typically installation of the AMS package is simple and straightforward. If you have problems installing it, contact us for assistance at firstname.lastname@example.org.
To install the AMS package you have to go through the following steps:
- 1. Decide which version to install
- 2. Download and install the software
- 3. Set up environment
- 4. Set up the license
- 5. Set up the scratch space
- 6. Test your installation
- 7. Configure ADFjobs queues and 3rd party software (optional)
Below we discuss each step separately.
Decide which version to install¶
Choose the released version or a snapshot. Normally, you will install the released version from the main download page. The bug-fixed binaries contains the most recent bug fixes. You may want to install a snapshot version if you know that some bugs have been fixed recently. The development snapshots page contains the latest developments. You will only want to download it if you need to use the latest functionality not available in AMS2019.
Choose a platform. AMS is available for a number of platforms. A platform is a combination of the processor architecture, operating system, MPI implementation, and any other specialties such as CUDA, forced AVX2 or OpenBLAS. Currently, the following platforms are officially supported and available from our website:
- Linux: x86-64 (64-bit): IntelMPI, OpenMPI, IntelMPI+CUDA (beta version only), IntelMPI+MKL with forced AVX2 (optimized for AMD Zen CPUs)
- Mac OS X Mavericks and later (10.9+): x86-64 (64-bit) with OpenMPI
- Windows: x86-64 (64-bit) with IntelMPI
32 vs. 64 bits. AMS2019 is only supported on 64-bit Operating Systems (and hardware). On 32-bit Windows ADF2017 can be used.
Choose MPI (on Linux). We advise to use the IntelMPI version if possible. It has been tested on both desktop and cluster machines and should work out-of-the-box on most cluster queueing systems. The IntelMPI runtime environment is distributed with AMS. If you prefer to use the OpenMPI version instead, keep in mind that if you wish to run multi-node calculations you need to use a local copy of OpenMPI 2.1 with support for your queueing system build in. The OpenMPI runtime environment is distributed with AMS but it is limited to single-node (or desktop) usage. Users of SGI should download a version specifically built for their platform, if available.
Cray XC. Cray users can use the IntelMPI version of AMS because it is binary-compatible with Cray MPI shared libraries. Read the MPICH ABI compatibility section in the Cray documentation for more information.
GPU acceleration: note: the CUDA version of AMS2019 will be made available at a later time. ADF can use NVidia GPUs for accelerating SCF cycles and frequency calculations. The GPU should have good double precision performance, compute capability 2.0 or higher and the operating system needs to be Linux with CUDA 8.0 installed. See the ADF manual on GPU Acceleration for details.
OpenBLAS (on Linux): AMS2018 introduced a new supported platform, optimized for AMD Zen (Ryzen/Threadripper/Epyc) processors. It uses OpenBLAS instead of the Intel Math Kernel Libraries (MKL) combined with AVX-2 optimized binaries. In AMS2019-3 this got replaced with forced AVX2 execution in MKL, as this gives better perfomance. The OpenBLAS version is still available on the bugfixed binaries page. This version will NOT work on older AMD processors.
Intel MPI + MKL for AMD-Zen (on Linux): AMS2019-3 introduced this combination to replace the OpenBLAS solution for AMD Zen (Ryzen/Threadripper/Epyc) processors. The binaries are compiled with AVX2 instructions, and the MKL library is forced to run in AVX2 mode. This currently gives the best performance on AMD Zen processors. This version will NOT work on older AMD processors.
Download and install the software¶
All systems: Download an installation package from one of the download pages mentioned above.
Unix and Linux: The installation package is a compressed tar archive. Untar it in the directory of your choice. Desktop users can follow the Linux Quickstart Guide to install AMS2019. Please note that in a cluster environment AMS must be installed on a shared file system accessible from all nodes of the cluster. The archive contains one directory, adf2019.xxx, which, when installed, will be referred to as $ADFHOME.
Mac OS X: The installation package is a disk image (.dmg) file. To install AMS on Mac OS X, double click on the downloaded disk image and drag AMS2019.xxx somewhere on your hard disk, such as the Applications directory. Be sure to read the enclosed ReadMe file for further important details!
Windows: The downloaded file is an executable Microsoft Installer package containing an installation wizard that will guide you through the installation process. Open the file after downloading, and follow the instructions. note: Do not close the console window that pops up during the installation! . Please note that you need Administrator privileges to install AMS. After or during the installer you might be asked by the Windows Firewall to grant network access to some executables, this permission is optional. For third-party firewall applications you will need to grant access from scmd.exe to the localhost.
Set up the environment¶
Windows users can skip to the next section since for them the environment is set up by the installation wizard.
Mac users may follow the UNIX instructions if they (also) wish to run from the command line. However, read the ReadMe file inside the dmg file for some important extra details! Alternatively, Mac users can start by double clicking the AMS2019.xxx application. The AMS2019.xxx application will automatically set the environment, and start ADFjobs for you. Next, all tasks (GUI or computational engines) started from this ADFjobs will inherit the proper environment.
For users of any Unix-like system the following step is mandatory.
For a single-user installation, the end-user is likely also the person installing AMS. If the user has a bash shell, it should be sufficient to source the $ADFHOME/adfbashrc.sh:
Alternatively, it is also possible to edit the $ADFHOME/Utils/adfrc.sh (for sh/bash/zsh users) or $ADFHOME/Utils/adfrc.csh (for csh/tcsh users) file to set a number of important environment variables and source the file:
or (for tcsh)
To set up the environment automatically when starting a new terminal, add the source command to the $HOME/.bashrc file. It is also possible to create a launcher icon for the GUI by running the $ADFBIN/create_linux_icon.sh script AFTER the environment has been set:
For a multi-user installation, you can either copy both adfrc.* files to the /etc/profile.d directory (after editing them) or, if your system supports modules, create a module file for AMS2019. The following environment variables must be defined in the module file:
- ADFHOME: AMS installation directory
- ADFBIN: should be equal to $ADFHOME/bin
- ADFRESOURCES: should be equal to $ADFHOME/atomicdata
- SCMLICENSE: complete path of the license file, typically $ADFHOME/license.txt
- SCM_TMPDIR: path to the user’s scratch directory, for example, /scratch/$USER. This directory must exist prior to the first execution of any program from the AMS package by that user. Thus it may be a good idea to add to the user’s profile a command that creates the directory in case it does not exist. You can also choose to use the same directory for all users. See SCM_TMPDIR Environment variable for more details.
A complete list of environment variables is provided in Appendix A.
If you are planning on using the GUI on a remote machine (for example via ssh with X-forwarding), make sure to also take a look at Using the GUI on a remote machine.
If you plan on running jobs directly from the command line, please do not mpirun the programs directly. The MPI startup is done automatically, for more information see the Additional Information on running MPI jobs.
Set up the license¶
Which functionality can be used on each computer is determined by the license file pointed to by the SCMLICENSE environment variable.
When you start any program from the AMS package (like ADF, BAND, ADFjobs or ADFinput) it will try to install a license file for you if necessary. To do this, some information is needed:
- Your e-mail address, this may be used to contact you and send you a license file.
SCM userid (uxxxxx):
- The same as the download login you should have received.
- The download password belonging to the SCM userid
- If checked, the SCM userid and password will be saved on your computer, so you do not have to enter them again. They are saved in a readable format, so only do this on a computer and account you trust.
Get license without user confirmation:
- If checked, a license request will automatically be issued when necessary, without user intervention. The intended use is for running on clusters, or for use in classroom/workshop situations. It will work only after arranging this with SCM first.
Click OK to request and install a license to run on your current machine. The information will be sent to the SCM license server.
If a license is available, or can be generated by the license server (for trial users) it will be installed automatically. Obviously you will need to have an internet connection with access to the outside world for this to work.
For Non-trial users, license requests are handled manually. After some time (between hours and days) you will be notified by mail that a license file has been prepared for you.
Run the software again (on the same machine with the same SCM userid), and in many cases the license will automatically be installed for you. If not, follow the “Manual license installation” instructions (further down on this page).
Click Cancel when you do not want to request a license to run on the current machine, if your machine has no internet connection, or if you wish to request and install a license file manually.
A window will appear (either a regular Text editor on your system, or the (License Update Status) appears telling you a license is needed:
It will show the info that needs to be mailed to SCM to get a license. You can copy that information to your clipboard by clicking the Copy Info button, or the usual copy tool in your editor.
Next, mail this information to email@example.com if you indeed wish to request a license for this machine.
After some time (hours-days as the license request is processed manually) you should receive a mail from SCM containing the license file.
You have receive a new license file via email. Do not make any changes to it, as that will break the license!
Mac users can normally just drop the license file on the AMS2019 application icon.
Other users should save the license file such that $SCMLICENSE points to it. Note the value of SCMLICENSE was shown in the License Update Status dialogue.
The default location of the license file (for Windows and the default SCMLICENSE value from a adfrc.* file) is set to $ADFHOME/license.txt, which means you should save the file as license.txt in the AMS installation directory. For macs the default location is “$HOME/Library/Application Support/SCM/license.txt”.
Unix-like users can generate the license information by running the following command at the shell prompt in a terminal window:
Output of this command from all computers on which AMS will be running, including all nodes of a cluster if applicable, must be sent to firstname.lastname@example.org.
After receiving this information SCM will prepare a license file matching your license conditions and e-mail it to you with instructions on how to install it.
After receiving your license file you will need to save it so that $SCMLICENSE points to it.
In a multi-user environment, make sure that permissions on the license file allow read access for everyone who is going to run AMS.
Note: floating licenses are only intended for Linux clusters.
If you have a floating license, you will need to follow these instructions below to set it up. The instructions are simple, but it is important you follow them exactly.
If you do not have a floating license you may skip this section.
Create a floating license directory
Make a new directory, for example FloatADF, to keep track of the running ADF processes.
This FloatADF directory must be:
- in a fixed location (the directory name should not change!)
- shared between / mounted on all nodes where you want to run ADF
- writable by all users that want to run ADF
- FloatADF must not be a subdirectory of $ADFHOME
cd /usr/share mkdir FloatADF chmod 1777 FloatADF
If you also have a floating license for other modules, you need to set up different directories and repeat this procedure for all these modules (for example FloatBAND, FloatReaxFF, FloatDFTB, …)
In the example, we have given all users read, write and execute permissions to this directory. If you wish to restrict this for security reasons, you may do so as long as all ADF users will have read, write and search permission for this directory. You may create an ADF user group for this purpose.
Important: the directory should not be a sub-directory of $ADFHOME as the directory name will change if you update to a new version! Also, do not put the license.txt file in the FloatADF directory.
The FloatADF directory may not be moved, deleted or renamed for the duration of your license because these actions will invalidate it!
E-mail us the license information Send the result of the following commands (using again the name FloatADF and the location /usr/share as an example) to email@example.com:
cd /usr/share ls -lid $PWD/FloatADF
Please note that output of the ls command must include the full path of the FloatADF directory.
Together with the output of the ls command above, you also need to send us output of the $ADFBIN/dirac info command from each computer on which ADF will possibly run, as the license file is still host-locked.
In the case of very large clusters, it is often sufficient if you send us the output of the $ADFBIN/dirac info command from the head node and 2 or 3 compute nodes. As most compute node names have the same beginning, we can then add them with a wild card (for example Linuxchem*). It is important when you use this facility, to let us know that the info you are sending are not all compute nodes. Otherwise the license will only run on these few compute nodes.
Set up the scratch space¶
Most programs from the ADF package use disk for temporary data. This data often takes a significant amount of space and is used frequently. To avoid run-time errors and performance loss you may want to make sure that the file system used for temporary files is both big and fast. The SCM_TMPDIR environment variable is used to tell the programs where to put their temporary data. Please note that SCM_TMPDIR should always be set. If it is not set then each process will create its own directory in the current working directory where is was started.
Please see Appendix A on additional information about the SCM_TMPDIR variable.
Using multiple disks
Sometimes, if you have multiple non-RAID disks in your system, you may want to spread scratch files across different physical disks for better performance. It is possible to request that every AMS MPI-rank creates its files in a different directory by adding “%d” in $SCM_TMPDIR. If a “%d” string is encountered in the value of SCM_TMPDIR variable it will be replaced by the MPI rank number of the process at run-time. This means, however, that you may have to create up to 128 or more (depending on the maximum number of processes in one parallel calculation) symbolic links on each node where AMS is supposed to run. You should also create a directory matching the SCM_TMPDIR’s value literally so that any process that does not interpret ‘%d’ could also run.
Example: suppose there are two scratch file systems, /scratch1 and /scratch2 located on two different physical disks of an 8-core workstation. We want the odd rank numbers to use /scratch2 and the even numbers to use /scratch1. One way to achieve this is to create an empty /scratch directory and create nine symlinks in it as follows:
ln -s /scratch1 /scratch/%d ln -s /scratch1 /scratch/0 ln -s /scratch2 /scratch/1 ln -s /scratch1 /scratch/2 ln -s /scratch2 /scratch/3 ln -s /scratch1 /scratch/4 ln -s /scratch2 /scratch/5 ln -s /scratch1 /scratch/6 ln -s /scratch2 /scratch/7
After that set SCM_TMPDIR to “/scratch/%d” and the ranks 0, 2, 4, 6 will use /scratch1 while ranks 1, 3, 5, and 7 will use /scratch2. When running AMS on a cluster it is better to combine multiple disks in a RAID 0 (striping) configuration as you probably do not want to create hundreds of symbolic links on each node.
Test your installation¶
This is a very important step and it should never be skipped.
Check the license file¶
Windows users test their license by double-clicking the makelicinfo.bat file in the AMS installation folder.
Unix users: first check that the license file has been installed correctly by running (at the shell prompt):
This should produce the output similar to the following:
Checked: /home/testadf/adf2019.xxx/license.txt License termination date (mm/dd/yyyy): 1/ 8/2019 According to it, you are allowed to run: ADF version 2019.990 ADFGUI version 2019.990 GUI version 2019.990 REAXFFGUI version 2019.990 BAND version 2019.990 BANDGUI version 2019.990 CRS version 2019.990 DCDFTB version 2019.101 DFTB version 2019.990 DFTBGUI version 2019.990 MOPAC version 2019.990 NBO version 2019.990 NBO6 version 6.000 QNDFTB version 2019.990 REAXFF version 2019.990 Utils version 2019.990 Number of procs you are allowed to use for: ADF : 1024 procs BAND : 1024 procs DFTB : 1024 procs ReaxFF : 1024 procs ===================================================== SCM User ID: u999999 release: 2019.101 :example.com: :Linuxmaster.example.com00:11:22:33:44:55: :ncores 12: :CPU Model Intel(R) Xeon(R) CPU E5-2630 v2 @ 2.60GHz: :DMY 1- 7-2016: :SRCID 3217288: ===================================================== LICENSE INFO READY
If the license check is successful (i.e. you get the “LICENCE INFO READY” message) you can move on. Otherwise check that the license file has been installed according to instructions above.
Run some basic tests¶
Verify that you can now run examples provided with AMS and that they give correct results. We recommend that you consult the Examples document for notes on such comparisons: non-negligible differences do not necessarily indicate an error. If you have a license for the GUI you can also run a few GUI tutorials as a test.
Note: the example .run files are complete Bourne shell scripts that should be executed as such, they are **not* input files to be fed into any program. The easiest way to run them is using ADFjobs.
Test the GUI¶
If the GUI is included in your license, check that you can start any of the GUI modules.
Enter the following command:
An ADFjobs window should appear.
Double click the AMS2019.xxx application to start it. An ADFjobs window should appear.
Double click the ADFjobs icon to start it. An ADFjobs window should appear. If the window does not appear or appears after a long delay then check the AMS-GUI requirements and check the firewall rules that should allow local communication.
All users should now be able to start ADFinput via the SCM menu on the top left of the menu bar.
Test parallel execution¶
It is very important to make sure that computer resources are utilized with the maximum efficiency. Therefore, you should check that each AMS job uses all processors/cores allocated to it and that the network is not overloaded with the disk I/O traffic.
Typically, when you submit a parallel job to the batch system you have a possibility to specify how many processors per node (ppn) to use. If the batch system you are using allows this, then make sure that you request ppn equal to the number of physical cores on each node. Note that AMD processors from the Familiy 15h based on the so-called “Bulldozer” cores have one floating-point unit (module) per two physical cores. On these Bulldozer architectures and its successors (Piledriver, Steamroller) AMS will probably perform best when you only run on half the physical cores. AMS tries to automatically detect the number of floating-point units (half the physical cores), but the user is advised to check this and otherwise set ppn accordingly. This is no longer the case for the AMD Zen (Ryzen/Threadripper/Epyc) architecture, but for these processors it is advisable to use the “Intel MPI + MKL for AMD-Zen” version of AMS2019-3. This version forces the Intel Math Kernel Library (MKL) into AVX2 execution mode for better performance, as MKL by default runs in a slower SSE2 mode on these AMD Zen processors.
It is also possible that your batch system does not allow you to specify ppn but instead it always assumes that there only one processor per node. In this case, you will need to edit the $ADFBIN/start file and add some commands for processing the hostfile.
In order to check that everything is working as intended, create a reasonably small AMS job and start it, preferrably on more than one node. After the job has finished, open the job’s .out file and find the table that looks like the following:
Parallel Execution: Process Information ============================================================================== Rank Node Name NodeID MyNodeRank NodeMaster 0 compute-0-0 0 0 0 1 compute-0-0 0 1 -1 2 compute-0-0 0 2 -1 3 compute-0-0 0 3 -1 4 compute-0-1 1 0 1 5 compute-0-1 1 1 -1 6 compute-0-1 1 2 -1 7 compute-0-1 1 3 -1 ==============================================================================
Check the names in the “Node Name” column and verify that the number of tasks per node is correct. If it is, then move on to the next section.
Test parallel performance¶
If the process allocation to nodes looks as expected then scroll down a bit to the following lines:
Communication costs MPI_COMM_WORLD: 1.026 usec per message, 0.0320 usec per 8-byte item Communication costs for intra-node: 1.023 usec per message, 0.0318 usec per 8-byte item
Make sure that you get reasonable numbers (less than 100 usec per message) both for intra-node and MPI_COMM_WORLD communication costs. Otherwise contact your system administrator. If only the MPI_COMM_WORLD value is large but the intra-node one looks reasonable, this may mean that the network link between machines is very slow and you may want to run single-node jobs only.
If all seems to be OK then move to the end of the file to the table called “Timing Statistics”. Before the table, there is a line of text beginning with “Total Used” that might look as follows:
Total Used : CPU= 8064.87 System= 1144.31 Elapsed= 9212.99
This shows how much time (in seconds) was spent in the AMS code (CPU) and in the kernel (System), and how long did it take for the calculation to complete (Elapsed). Ideally, the system time should be small compared to the CPU time and the latter should be close to the Elapsed time. The system time will not always be a small number, however, the sum of the System and CPU times should always give a number very close to the Elapsed time. If this is not the case then it means that AMS has to spend a lot of time waiting for something. This can be, for example, disk I/O or network communication.
If you notice that the system time portion is enormously large or that there is a large difference between the CPU+System and the Elapsed time, then repeat the job with the “PRINT TimingDetail” keyword in the input and contact the SCM support.
Configure ADFjobs queues and 3rd party software (optional)¶
If you would like to use the GUI as a front-end to submit your jobs to queues, you should configure those queues in ADFjobs. Third party software MOPAC and ffmpeg (export movies) may also be installed separately.
A video shows how to set up remote queues on Windows, other platforms are even easier.
ADFjobs can submit and monitor jobs for you on different queues, both locally and remotely. You can use the GUI on your desktop or laptop for creating, submitting and analyzing your jobs, while running the jobs on remote compute clusters. In that case you should establish and ssh-agent connection, to enable remote job managing by the GUI. If you just run local jobs without a queuing system, skip this section.
To set up a new batch queue, select Queue → New… → and choose an example queuing system to starting with (LSF, PBS, or SGE). Modify the input files such that they correspond to your queue configuration:
- change the Name of the queue
- set the Remote host to the hostname of your cluster
- set Remote user to your username on that cluster
- change the Run command in accordance with your typical queue set up. The $options corresponds to the input box that can be modified in ADFjobs before submitting, e.g. the time or number of nodes
- you can set what $options defaults to in the Default Options field
- change the Kill, Job status, System status commands, if necessary
- you may define a Prolog or Epilog command
Managing remote jobs¶
To manage remote jobs, you need automatic authentication with an ssh-agent via an ssh key pair. This is most easily set up on MacOS or Linux, but is a bit more involved for Windows.
ssh connection on MacOS and Linux
If you don’t have an ssh key pair already, you can generate one in a terminal: ssh-keygen -t rsa Put your pass-word protected public key on the compute cluster(s) in ~/.ssh/authorized_keys.
Next, you should establish a connection via an ssh-agent. On MacOS an ssh-agent runs by default, and with keychain you just have to type your password once when you make the first connection.
On various Linux installations ssh-agent is also running by default. If an ssh-agent daemon is not yet running in the terminal where you will run ADFjobs, start the daemon, e.g. by adding eval $(ssh-agent bash) to your .bashrc. You need to add your private key to the ssh-agent daemon: ssh-add.
ADFjobs will use SSH to connect to the remote systems, making a new connection for everything it does. This means that there will be many connections to the remove machine(s). If you are using OpenSSH (both on the local and the remote machine), you can instead use one ssh connection and keep it alive, reusing it for many things. This makes ADFjobs faster, and may avoid complaining system administrators of the remote machines.
To do this, set the environment variable SCM_SSH_MULTIPLEXING to yes (for example via the GUIprefs application).
ssh connection on Windows
PuTTY tools, automatically installed with ADF, can be used to set up an ssh-agent connection. Follow these steps, using the PuTTY programs located in the bin\Putty directory in your AMS installation directory.
- Run puttygen.exe, set the key length to at least 2048 bits and press the Generate button. This will generate an ssh key pair. Type a key passphrase and save the private key to a file. Add the corresponding public key to the ~/.ssh/authorized_keys file on the remote host.
- Run pageant.exe and add your private ssh key (right-click on the Pageant icon in the task bar) with the correct passphrase.
- Open ADFjobs menu Help|Command-line and run “ssh user@host uptime” using correct username and hostname. Type yes if prompted to store the key in cache. If you get prompted for a password or get authentication errors then something is not configured correctly. The remote job submission will not work until all problems are resolved.
Now you can close the prompt and start using ADFjobs to manage remote jobs. You may test your queue configuration: Jobs → Generate Test Job and assign it to your new queue before running the test job.
Centrally defined queues
A system administrator may also define a queue centrally, which may then be picked up by any user automatically via Dynamic queues. Consult the GUI manual for information on how to set these up in ADFjobs.
The ADFmovie GUI module can make MPEG videos of the system as you see it in the ADFmovie window. This can be an MD trajectory or vibrational modes, or the course of geometry optimization. For this to work, you need to install the free ffmpeg program from the FFmpeg website and make sure it can be found in the path. The simplest way to do this is to copy the ffmpeg executable to $ADFBIN.