Galileo temporary stop on March 24th
- /
- HPC Center news
- /
- Galileo temporary stop on...
Dear Users,
we are glad to inform you that Galileo cluster is entering the production phase for the Xeon Phi accelerators. As already announced, the cluster is equipped with two Intel Phi 7120p per node on 384 nodes, which will now become available for production.
In order to manage the scheduling and the execution of the jobs on the Xeon Phi accelerators, an update of the PBS version is in order. This update will also strengthen the management of the resources (cpus, memory, accelerators etc.) assigned to the jobs, to guarantee their reliable use.
The update requires the stop of the compute nodes next March, 24th. The job scheduler will be set up to ensure that only jobs, which will end by 9:00 a.m. of March 24th, will start running in the next hours. For such reason, while no jobs will be killed at the start of the operations, jobs requiring a longer walltime will have to wait until the end of the procedure before being considered for scheduling. The login nodes and storage areas will instead remain available.
We will inform you at the end of the installation, which is expected to be complete in the early afternoon. Please, refer to the relevant section of our on-line guide to Galileo, http://www.www.hpc.cineca.it/content/galileo, for a brief guide on the production and programming environment for the Xeon Phi.
Best regards
HPC User Support @ CINECA