We just got this bad boy for AI research

Polish Virgo group at CAMK has just received a new high performance system for deep learning, a DGX-like node manufactured by PNY with 8 NVIDIA Tesla V100 SXM2 modules and NVLink. The server is equipped with 2 16-core CPUs, 768 GB RAM and 100 Gbps Infiniband network interface. The total performance of GPU modules is almost 1 petaflops (GPU FP16, half-precision)!

The purchase was possible with the MNiSW grant for expansion of the strategic IT infrastructure for science. The Polgraw team is a member of international consortium Virgo working on analysis of data from interferometric detectors of gravitational waves within LIGO-Virgo-KAGRA collaboration. The new system will be used mainly for searching for astrophysical signals and their parameter estimations, as well as to research detector noise using deep neural networks (CAMK belongs to the international Virgo consortium working on analysis of data from interferometric detectors of gravitational waves within LIGO-Virgo-KAGRA collaboration).