Scientific Computing
Genesis Cluster
Scientific coordinator of the
Genesis Cluster, the HPC infrastructure of the INAF national program
Arxes hosted at INAF-IAPS.
The Genesis Cluster provides 144 physical cores distributed over two SuperMicro computational nodes based on Intel Skylake architectures and three PowerDell computational nodes based on AMD Epyc architectures, complemented by a SuperMicro storage server based on Intel Skylake architecture hosting the shared filesystem of 60 TB. All servers have 16 GB of memory per core and are connected at 10 Gbps through a dedicated switch.
The disk array of the storage server is configured as RAID 6 with two additional hot spares disks. An additional disk array of 8 TB configured in RAID 6 is hosted in one of the PowerDell computational nodes and is used to backup the home directories of the users. Further redundancy is provided by nightly backups on the INAF GDrive cloud.
HPC, programming and IT knowledge
- High-performance computing (HPC): parallel programming (OpenMP standard), GPGPU programming (OpenACC standard), distributed computing (data parallel programming), software profiling and optimization for scientific computing
- Programming styles: object-oriented programming, mixed-language programming with Fortran-C/C++, procedural programming
- Programming and scripting languages: Fortran, C/C++, Python, Bash, Matlab.
- Scientific plotting languages: Gnuplot
- Operating systems and system administration: Linux, UNIX, Windows.
Released Software
- DPI (ascl:1504.012): FORTRAN77 library that supplies the symplectic mapping method for binary star systems for the Mercury N-Body software package (ascl:1201.008) developed by J. E. Chambers. The binary symplectic mapping is implemented as a hybrid symplectic method that allows close encounters and collisions between massive bodies and is therefore suitable for planetary accretion simulations. The current release only supports S-type binary star systems. Released under GPL3 license. Download
Webmaster: Mirko Riazzoli