High Performance Computing Center

Scuola Normale Superiore

March 29, 2022 - JupyterHub improvements

A new version of our customized JupyterHub installation is now available!
Release notes:

  • Notebooks can now be run on any compute node you have access to.
  • You can now provide a bash script to be executed before starting the notebook server, e.g. with the purpose of initializing the environment.
  • Several usability improvements and bug fixes.

Please report any problem to hpcstaff@sns.it.

February 8, 2022 - New compute nodes: Anacreon and Cinna

Two new groups of nodes has been added to Trantor, "Anacreon" and "Cinna".
Technical specifications are listed here.

Access to Anacreon and Cinna nodes is restricted respectively to members of the "bioinfo" and "compnanobio" groups.

Use the corresponding q07anacreon, q14anacreon, q07cinna and q14cinna PBS queues to submit jobs to these nodes (see here for details).

January 13, 2022 - Cluster stop from January 31 to February 2

Due to a maintenance intervention on the cooling system in San Silvestro, there will be a cluster shutdown from Monday January 31 10:00 AM to Wednesday February 2.

Please be sure to terminate all jobs and recover any data you may need before Monday January 31 10:00 AM.

December 17, 2021 - Important limitations of the storage system's snapshotting feature

The native snapshotting mechanism our storage system does not support the preservation of hard-links in snapshots. That means that if multiple files points to the same blocks of data on disk, only one of those files will be preserved in snapshots (you can find a gentle introduction to hard-links here).

That fact implies that, in those scenarios where hard-links are in use, there may be loss of information when recovering files from snapshots (since only one hard-link for each data file would be recovered). Examples of such scenarios include:

  • Software installations that make use of hard-links (such as conda virtual environments).
  • Hard-links manually created by users.
  • Files generated as output by applications, if the software creates hard-links (e.g. as a way to avoid data duplication). Fortunately, such cases are quite rare.

Unfortunately, there is nothing we can do to overcome this limitation, so please avoid the use of hard-links as much as possible and use soft-links instead.

October 11, 2021 - New head node with remote desktop

A new head-node equipped with the Xfce desktop environment is now available: trantor03.sns.it

You can connect to the new head-node via ssh (as usual) or by opening a remote desktop session. Detailed instructions are discussed in the wiki.

September 15, 2021 - New compute node: Kalgan01

A new node (DELL R750) has been added to Trantor. It is named kalgan01 and equipped with 2 Intel Xeon Platinum 8380 CPU @ 2.30GHz (40 physical cores per socket), 2 TB of memory and a scratch area of 8.7 TB.

The corresponding q07kalgan e q14kalgan PBS queues have been created.

Access to Kalgan01 is restricted to members of the "kalgan" group .

August 16, 2021 - New Aurora nodes

A group of 11 new nodes (HPDL650 Gen10) has been added to Trantor. They are named aurora[01-11] and are equipped with 4 Intel Xeon 6252N @2.3 GHz (24 physical cores per socket), 512 GB of memory and a scratch area of 4.9 TB (with 750 GB of SSD cache).

The corresponding q07aurora e q14aurora PBS queues have been created.

Access to Aurora nodes is restricted to members of the "diamond" group .

April 12, 2021 - Request of creation of a storage area destined to a research project

In the 'Forms' page you can find a form to request, to the Committee and the Staff, additional storage for a specific research project. Such additional storage will be reserved for a limited amount of time (max 1 year).

The creation of a new area must be proposed by tenured personnel and must include the list of users that can share read/write access to the area.

You can download the form here.

March 18, 2021 - Changes to PBS queues settings

We changed the settings related to the maximum number of running and queued jobs (per user) accepted within a queue, as follows:

  • q07daneel: max 2 running jobs and 8 queued jobs per user.
  • q14daneel: max 2 running jobs and 8 queued jobs per user.
  • q07hal: max 1 running job and 4 queued jobs per user.
  • q14hal: max 1 running job and 4 queued jobs per user.
  • q07helicon: max 4 running jobs per user. No limits on queued jobs.
  • q14helicon: max 4 running jobs per user. No limits on queued jobs.
  • q07artes: max 3 running jobs per user. No limits on queued jobs.
  • q14artes: max 3 running jobs per user. No limits on queued jobs.
  • q07diamond: max 10 running jobs per user. No limits on queued jobs.
  • q14diamond: max 7 running jobs per user. No limits on queued jobs.

March 8, 2021 - New Intel compilers suite available on Trantor

The Intel OneAPI compilers suite is now available on Trantor. You can use it by loading the module intel/2021.1.1 and any relevant module under intel/*.

The licence is "single fixed Multi-Node", meaning that only one person at a time can use the compiler, from any computer.

Since the suite has many tools and libraries, we decided to install only some of them, to avoid cluttering. Please contact us if you need something that is not installed.

Dec 23st, 2020 - Singularity is now available on daneels and hals

Singularity Community Edition is now available on the Daneels and Hals nodes of the Trantor cluster.
For details, please refer to the dedicated section of the wiki.

Dec 23st, 2020 - Daneel01 temporarily back in the queues

Daneel01 has been put back in the queues; a few caveats on this node:

  1. GPU0 is detected but does not work correctly, so be sure not to use it in your applications.
  2. We started a maintenance call on this; likely the replacement will be available in january 2021, but in any case consider that we may be forced to shut it down with very short notice.

Dec 21st, 2020 - Hal nodes

The hal01 and hal02 nodes (Quad socket Dell R840 with 56 core Xeon Platinum 8280, 3.0 TB of memory and 9 TB of scratch) are now available in the Trantor cluster; the q02hal and q07hal queues have been created.
Note that even if these nodes have a lot of powerful cores scaling and running multiple jobs has not been up to expectations, so consider the current queue and node setting experimental.

Dec 7th, 2020 - artes MPI nodes

A new group of nodes, called artes0[1-6] (dual Xeon(R) Gold 5218 CPU @ 2.30GHz, 32 cores, 128 GB memory ) is now open to computation in the Trantor cluster, for people in the diamond group, set up for MPI type jobs.

Dec 1st, 2020 - R support in JupyterHub

JupyterHub@Trantor (https://jupyter.sns.it) now supports R-based notebooks.
In order to exploit this feature, you have to create an R-based Conda environment as explained in the updated User Guide.
In addition, the "Brief introduction to Conda" has also been updated with instructions on creating R-based environments.
As before, the access to JupyterHub@Trantor must be explicitly enabled, so if you are interested send a request to hpcstaff@sns.it.