| Scientific computing systems | <img alt="" src="http://www.cira.it/PublishingImages/immagini-home/IMG_1353.JPG?RenditionID=8" style="BORDER:0px solid;" /> | https://www.cira.it/en/research-infrastructures/scientific-computing-systems/Scientific computing systems | Scientific computing systems | <h3>Description and operational capabilities</h3><p>CIRA ensures the availability and the operation of Scientific Computing Resources needed to perform all the activities that require an intensive use of high performance computers. </p><p>The only access to the high-performance computer is not enough to achieve high levels of performance. For this reason, in order to exploit all the potential hardware capabilities, a CIRA staff highly specialized in the parallel and numerical computing fields is available. The work of this staff concerns: the optimization of codes, the rewriting of inefficient software, the development of efficient algorithms and the benchmarking of new hardware and tools. For this scope, large investments on activities of study and research, in cooperation with universities and research centres, are carried out. </p><p style="text-align:justify;">The main operational capacities are listed below :</p><ul><li><p>analysis of CIRA high performance computing needs and architecture definition of the systems to be acquired</p></li><li><p>management and maintenance of the supercomputing systems installed</p></li><li><p>user support for optimal use of compilers and tools</p></li><li><p>management of the catalogue of the scientific software available at CIRA, both commercial and developed in house</p></li><li><p> development and / or optimization of numerical algorithms </p></li><li><p> research and assessment of new hardware and software technologies for HPC</p></li></ul><h3>Technical features</h3><p>At the time, the computer center consists of some multi -node multiprocessor Cluster HA, for a total of 56 Nodes dual-socket of Intel Xeon processors. Moreover, some Nvidia M2090 GPU are available , for a total of 744 cores. The aggregate power is equal to 18 TFlops (including the GPUs ), with a 3.2 total memory of 3.2 TBytes. The connection among nodes is based on the InfiniBand technology. Within the first months of 2017 a new system, having an aggregate power of at least 50 TFlops , will be installed. During the 2018 this system will be successively upgraded to 300TFlops also by means of new generation processors.</p><p>For what concerns the application software, many commercial codes ( eg . ANSYS , NASTRAN ) and in house computer programs (see the catalogues) are available. </p><h3>Partners</h3><p>Research and experimental activities are also carried out in collaboration with the "Federico II" university, the Second University of Naples (SUN), ENEA and INFN, CNR.</p> | | <p>CIRA ensures the availability and the operation of Scientific Computing Resources needed to perform all the activities that require an intensive use of high performance computers.</p><p>The only access to the high-performance computer is not enough to achieve high levels of performance. For this reason, in order to exploit all the potential hardware capabilities, a CIRA staff highly specialized in the parallel and numerical computing fields is available. The work of this staff concerns: the optimization of codes, the rewriting of inefficient software, the development of efficient algorithms and the benchmarking of new hardware and tools. For this scope, large investments on activities of study and research, in cooperation with universities and research centres, are carried out.</p><p style="text-align:justify;">The main operational capacities are listed below :</p><ul><li><p>analysis of CIRA high performance computing needs and architecture definition of the systems to be acquired</p></li><li><p>management and maintenance of the supercomputing systems installed</p></li><li><p>user support for optimal use of compilers and tools</p></li><li><p>management of the catalogue of the scientific software available at CIRA, both commercial and developed in house</p></li><li><p> development and / or optimization of numerical algorithms</p></li><li><p> research and assessment of new hardware and software technologies for HPC.</p></li></ul> | <p>At the time, the computer center consists of some multi -node multiprocessor Cluster HA. The main system, named "Turing", consists of 42 computational nodes based on INTEL E5-2697v4 processors, for a total of about 1500 cores and some GPU nodes. It reachs an aggregate power of 60 Tflops. This cluster was installed on May 2017. Other small clusters are available, for additional 744 cores. The aggregate power available at CIRA is about 80 TFlops (including the GPUs ). At the and of 2018 the "Turing" system will be upgraded to 300TFlops also by means of new generation processors.</p><p>For what concerns the application software, many commercial codes ( eg . ANSYS , NASTRAN ) and in house computer programs (see the catalogues) are available.</p> | | <p>Research and experimental activities are also carried out in collaboration with the "Federico II" university, the Second University of Naples (SUN), ENEA and INFN, CNR.</p> | |
