CRESCO - Computational RESearch centre on COmplex systems

HOME| CRESCO| Guides/Tutorials| Contacts| F.A.Q.

CRESCO: System Access

Access to frontend nodes via SSH
Access via the ENEAGRID FARO web interface

1. Access to frontend nodes via SSH

Computing resources in ENEAGRID are distributed over six ENEA centres: Portici, Frascati, Casaccia, Brindisi, Trisaia and Bologna. In particular, the clusters installed at Portici, Frascati and Casaccia, represent the main computing facilities, while the centres of Trisaia and Bologna host AFS storage resources.

ENEA-Portici hosts the largest quota of HPC systems. Our supercomputers fulfill the needs of ENEA researchers and external partners. They are used to solve problems in a plethora of computational research fields.

The following links provide descriptions of the CRESCO HPC systems installed at ENEA-Portici:

To access all frontend nodes of the clusters, the user may login via SSH to the super-frontend

cresco-in.portici.enea.it

Frontend nodes serve as control servers to launch applications via LSF or SLURM, to edit the user's own launch scripts, to compile codes and - as is the case for dedicated frontends (see below) - to make short test runs. All computing nodes can be reached only via LSF or SLURM. For a short tutorial on the basic commands to submit batch jobs go HERE for LSF or go HERE for SLURM.

The section Batch Jobs / LSF of the present document contains specific information on CRESCO.

The section Batch Jobs / SLURM of the present document contains specific information on CRESCO.

Please find below the list of frontend nodes (the domain suffix ".portici.enea.it" is omitted, e.g. the hostname of the first frontend node of CRESCO4 is cresco4x001.portici.enea.it):

Section 4 of CRESCO @ Portici
cresco4x001
cresco4x002 (use it for interactive runs as well)

Section 6 of CRESCO @ Portici
cresco6x001
cresco6x002 (use it for interactive runs as well)

Section 7 of CRESCO @ Portici
cresco7x001
cresco7x002 (use it for interactive runs as well)

Section 8 of CRESCO @ Portici
cresco8x001
cresco8x002 (use it for interactive runs as well)

Special section with large memory nodes @ Portici
cresco4sm01 (12/24 physical/logical cores, 768 GB RAM, use it for interactive runs as well)
cresco8-hbm01 (56 physical cores, 1TBB RAM, use it for interactive runs as well)
cresco8-hbm02 (56 physical cores, 1TBB RAM, use it for interactive runs as well)

Special GPU Section of CRESCO @ Portici
cresco4sk1 cresco4skp1 (Tesla K40i dismesse)
cresco8-gpu01 cresco8-gpu02 (Intel Max Pontevecchio)

CRESCO @ Casaccia
crescoc1

CRESCO4F @ Frascati
cresco4fx001

The super-frontend cresco-in.portici.enea.it is a bridge towards physical cluster frontends. This node is a virtual machine without computing resourses. It should not be used as a physical frontend.

Frontend nodes allowing interactive runs are the only ones that users should use to perform (even parallel) test runs. On these nodes we permit test runs consuming even considerable fractions of the node computing resources. Test runs performed on other frontends are killed by system admninistrators without notice. Remind that test runs can be useful to test compilation, optimization, debugging and so forth. Test runs should be always performed before submitting a real job to the LSF scheduler.

Frontend nodes allowing interactive runs have a different memory management, that allows a more intensive use of the node resources with respect to standard frontend nodes. This is the reason why test runs are only permitted on the above-specified frontend nodes..

[contents]

2. Access via the ENEAGRID web interface FARO

Users can access all clusters via the FARO web interface, by connecting on their browser to the website

cresco-in-gui.portici.enea.it.

[content]

CONTENTS