TeraStat2 (TS2) is the general-purpose supercomputing cluster of the Department of Statistical Sciences for solving mathematical and statistical models on Big Data. It consists of 12 computing nodes, running under Linux and each equipped with 2 AMD Epyc 7452 processors and 256GB of RAM, and 12 computing nodes equipped with two 8-Core Intel Xeon E5-2630 processors (16 virtual cores) and 64GB of RAM. One of the latter is equipped with 512GB of RAM. The system has a total of 1,920 cores. The cluster also has 76TB of shared storage and an InfiniBand QDR fibre-optic network, which supports 40Gb/s bandwidth between nodes and between nodes and storage.
Modes of use
Being a high-capacity system, the infrastructure is accessiblee through the execution of batch jobs. User prepare a description of the experiment in a particular format, together with all necessary material (e.g. input data). This description is sent to a queuing system, running on TS2, which keeps track of the computing resources consumed by each user. As soon as there are sufficient resources, the experiment requested by the user is automatically put into execution. The user will be notified via e-mail at the end of the job and will be able to reconnect to TS2 to get the results.
TS2 usage is free of charge to all Sapienza staff (professors, researchers, PhD students, research fellows) for developing projects involving a high-performance scientific computing component. For particularly intensive use, a non-commercial value subscription is required, which will contribute to covering the costs of maintaining and updating TS2. This subscription allows exclusive access to a fixed amount of computing resources and is valid for one year. Different types of subscription are available, with different amounts of allocated resources and associated accounts.
Click here qui to request a free account
TS2 staff takes care of the installation, maintenance and updating of a set of software packages and libraries of common interest to general users. Users needing software and libraries not initially present on TS2, may install them locally to their user accounts or request their installation to the staff if software is considered to be of general interest. In any case, only open source or licensed software may be installed.
TS2's main source of funding was the medium-sized university equipment project entitled 'TeraStat2: high-density supercomputing infrastructure for solving high-complexity data-centric problems' (code no. MA31916B8891453A), for which Prof. Umberto Ferraro Petrillo is the scientific head, financed in 2019 by the amount of 55k euro. It was joined by several co-funding contributions from the Department of Statistical Sciences including those provided by the Master degree in Big Data. Statistical Methods for the Knowledge Society and from the Master Degree in Data Intelligence and Decision Strategies.
TeraStat2 is the evolution of the TeraStat cluster, initially set up by the Department of Statistical Science in 2011. Further information about TeraStat is available here.