Besso¶

Besso is a small Alps cluster that provides development resources for porting software for the SKA project. It is provided as is, without the same level of support as the main platform clusters.

Storage and file systems¶

Besso uses the HPCP filesystems and storage policies.

Getting started¶

Logging into Besso¶

To connect to Besso via SSH, first refer to the ssh guide.

~/.ssh/config

Add the following to your SSH configuration to enable you to directly connect to besso using ssh besso.

Host besso
    HostName besso.vc.cscs.ch
    ProxyJump ela
    User cscsusername
    IdentityFile ~/.ssh/cscs-key
    IdentitiesOnly yes

Software¶

uenv¶

Besso is a development and testing system, for which CSCS does not provide supported applications.

Instead, the prgenv-gnu programming environment is provided for the both the a100 and mi200 node types.

Containers¶

Besso supports container workloads using the Container Engine.

To build images, see the guide to building container images on Alps.

Running jobs on Besso¶

Slurm¶

Besso uses Slurm as the workload manager, which is used to launch and monitor workloads on compute nodes.

There are multiple Slurm partitions on the system:

the a100 partition contains NVIDIA A100 GPU nodes
the mi200 partition contains AMD Mi250x GPU nodes
the normal partition contains all of the nodes in the system.

name	max nodes per job	time limit
`a100`	2	24 hours
`mi200`	2	24 hours
`normal`	4	24 hours

Default MPI plugin

The default MPI plugin on Besso is cray_shasta. This plugin is correct for applications that use Cray MPICH, for example applications in uenv. Applications that use OpenMPI or MPICH must set --mpi=pmix or --mpi=pmi2.

See the Slurm documentation for instructions on how to run jobs.

FirecREST¶

Besso can also be accessed using FirecREST at the https://api.cscs.ch/mwa/firecrest/v2 API endpoint.

Maintenance and status¶

There is no regular scheduled maintenance for this system.