1. GRID COMPUTING

DFCTI develops and manages two grid sites, RO-07-NIPNE and GRIDIFIN. RO-07-NIPNE is dedicated to the computational support of the ALICE, ATLAS and LHCb experiments, within the WLCG collaboration. For this purpose, in 2018 RO-07-NIPNE offered to WLCG more than 4,400 CPU cores and 2,300 TB storage space, ranking first at national level regarding the resource capacity.
GRIDIFIN supports the ELI-NP project (to which it provided 546,000 CPU hours in 2018) and also serves other research communities in physics and related areas.

2. HIGH-PERFORMANCE COMPUTING

Since 2006, DFCTI's staff has developed the high-performance computing (HPC) infrastructure of IFIN-HH, consisting of parallel clusters in Gigabit Ethernet, Myrinet and Infiniband communication technologies. Today, heavier MPI computations are performed on the IBM_BC cluster (IBM/Lenovo BladeCenter), on more than 1,350 CPU cores and 40,000 GPU cores.

3. CLOUD COMPUTING

Starting in 2014, DFCTI developed a cloud copmuting cluster, CLOUDIFIN, which is registered in the EGI infrastructure.

Northern view of the RO-07-NIPNE cluster	The data center that hosts the grid storage

Network Infrastructure

DFCTI hosts the point of presence (PoP) through which the Magurele Platform of Physics is connected to the Romanian NREN, RoEduNet, at 30 Gigabits/sec. Due to the increasing grid traffic, DFCTI is currently implementing a plan to gradually increase the bandwidth of IFIN-HH's connection up to 100 Gigabits/sec.