| Processors | 3.920,0 | |
| RAM | 7.840,0 | GB |
| Power | 72.030,0 | GFLOPS |
Magerit is a cluster consisting of 245 computer nodes.
To run jobs in the supercomputer you must request an account using the project and account request form. The Magerit users manual (in spanish only).
Magerit is designed to work with batch processes. All jobs are submitted to the resources manager and scheduler SLURM, in order to optimize Magerit occupation.
Setting maximum existing queues:
| QoS | Nº CPUs | WallClockLimit | Priority |
|---|---|---|---|
| debug | 16 | 00:10:00 | – |
| standard | 512 | 72:00:00 | Low |
| partner | 1024 | 72:00:00 | Medium |
| premium | 1024 | 72:00:00 | High |
| class_a | 1024 | 72:00:00 | High |
| class_b | 1024 | 36:00:00 | Medium |
| class_c | 512 | 24:00:00 | Low |
Magerit is a cluster consisting of 260 computer nodes, of which 245 nodes are eServer BladeCenter PS702 2S with 16 Power7 processors 3'3 GHz (26.4 GFlops) and 32 GB de RAM, and the rest are 15 nodes eServer BladeCenter HS22 with eight Intel Xeon 2'5GHz (10.2 GFlops) processors with 96 GB RAM, implying 4,160 CPUs and 9.2TB RAM. All the nodes operate independently and with the same software configuration.
El sistema dispone de una capacidad de almacenamiento local de unos 192 TB, proporcionado por 256 discos de 750 GB, que utiliza un sistema distribuido y tolerante a fallos (GPFS).
La conexión exterior se realiza a través de RedIRIS mediante enlaces de 1Gb y 10 Gb. Está prevista la actualización de esta conexión según lo aconseje la demanda.
All nodes have installed SLES11SP1 (kernel 2.6.32). Besides all the standard working environment on a Linux system, provides many tools and libraries for scientific computation.
Others softwares and libraries are available by request for support center.
Madrid Supercomputing and Visualization Center.
Users