ADF modeling suite - Hardware and Software Requirements
The amount of memory you need depends greatly on the kind of calculation you perform. For most calculations 256 MB per CPU will be sufficient, but if you have lots of memory available most program will run significantly faster. It also reduces the amount of disk I/O, which may speed up your calculation depending on the I/O sub-system and the operating system.
Memory requirement grows with the system size. For example, for a molecule containing 100 atoms with a DZ basis set it may be sufficient to have 256 MB per CPU but for a molecule with 1000 atoms up to 8 gigabytes per CPU may be required. Also, if you are going to perform TDDFT, relativistic spin-orbit or analytical frequency calculations then the amount of memory should be larger. As an indication, an analytical vibrational frequency calculation of a organometallic complex containing 105 atoms with a TZP basis set uses up to 1GB of RAM per process but it can be done with less, even though not as efficiently.
For the installation of the package on Linux/Unix you need from less than 256 MB (without sources) to 1 GB (with sources and compiled objects). On Windows, about 900MB of disk space is required for ADF installation. The (scratch) disk space requirements for running ADF greatly depend on what type of calculation you perform. For most ADF calculations 1 GB of disk space will be enough. For BAND calculations you may need from a few to up to a few tens of gigabytes (for large calculations) of free disk space.
- Supported platforms:
Cray XE (x86-64, CLE).
Cray XT (x86_64,CLE).
IBM AIX (ppc64).
Itanium2 Linux (ia64).
MacOS X 32-bit (Intel).
MacOS X 64-bit (Intel).
PC Linux 32-bit (i386).
PC Linux 64-bit (x86-64).
PowerPC Linux 64-bit (ppc64).
SGI Altix (ia64).
SGI Altix ICE (x86-64).
Windows 7, Vista (64-bit).
Windows 7, Vista, XP (32-bit).
First of all, on the most popular systems (Linux, MacOS and Windows) a network card must be present in the computer as its hardware MAC address is used as the computer's ID for licensing purpose. Besides that, when using Platform-MPI, the computer must be connected to a network and have a valid IP address and the address must not change while an MPI job is running. This requirement might be lifted in a future version of Platform MPI.
As far as performance concerned, a switched (Gigabit or Fast) Ethernet network is typically sufficient for good parallel performance on up to eight nodes. If you are going to use more nodes you may need faster communication hardware, such as Myrinet or Infiniband, to get the best performance.
All commercial parallel machines (SGI, IBM, HP, …) have very good communication facilities, so the package should run fine on those platforms as far as the communication system is concerned.
If you have a license for the source code, you can compile and link the source yourself, with or without your own modifications.
The source consists mainly of Fortran95 code, with some small parts written in C. You need to use the same Fortran and C compilers as we are using on the same platform, since some parts of the code are available only as object modules.
Fortran90 compiler: Intel Fortran, XLF.
C compiler: Microsoft Visual C++, GCC, Intel C/C++, XLC.