<p>30/12/2019<br />
Dear Users,</p>
<p>This is to inform you that the installation of the new accelerated partition<br />
of CINECA (Marconi100) proceeds as scheduled. The new racks have been<br />
delivered to CINECA and will be installed soon. We remind you that the old<br />
MARCONI-A2 (KNL) partition will be out of production within the end of<br />
January 2020 and Marconi100 will start its production phase in March 2020.</p>
<p>MARCONI100, as announced, is based on the IBM Power9 architecture with NVIDIA<br />
Volta GPUs. Specifically, each node will host 2x16 cores IBM POWER9 AC922<br />
at 3.1 GHz with 256 GB/node of RAM memory and 4 x NVIDIA Volta V100 GPUs per<br />
node, Nvlink 2.0, 16GB. The number of nodes will be 980, totallying<br />
31360 cores. Internal Network: Mellanox Infiniband EDR DragonFly+</p>
<p>To help users porting their own code on this new GPU accelerated<br />
architecture, the 2020 updated catalogue of available training courses,<br />
including CUDA programming, is available and can be consulted here<br /><a href="https://eventi.cineca.it/en/hpc">https://eventi.cineca.it/en/hpc</a></p>
<p>Best Regards,</p>
<p>HPC User Support @ CINECA</p>