Scheduler Defaults
Max walltime | 2 hours |
Default walltime (if not specified) | 15 minutes |
Max running jobs per user | 2 |
Max queued jobs per user | 20 |
Max cores per job | 16 |
Max compute per user | 100 CPU hours |
Max GPUs per job | 2 |
Beta is intended only for non-commercial, academic research and instruction, and no sensitive data (PHI, HIPAA, FERPA, etc.) may be stored or processed there. It is suited to testing scripts and is not intended for production-level research.
VIEW ALL POLICIESData is not backed up. You are responsible for long-term storage of data.
VIEW STORAGE OPTIONSTo use Beta, you must establish a Flux user account.
Fill Out FormContact arcts-support@umich.edu for help.
EMAIL SUPPORTMax walltime | 2 hours |
Default walltime (if not specified) | 15 minutes |
Max running jobs per user | 2 |
Max queued jobs per user | 20 |
Max cores per job | 16 |
Max compute per user | 100 CPU hours |
Max GPUs per job | 2 |
Beta is a preview cluster that allows users to become familiar with SchedMD’s Slurm Workload Manager, which will be used in our upcoming Great Lakes cluster. All standard compute nodes in the Beta cluster have 16 core Intel(R) Xeon(R) CPU E5-2670 processors at 2.60GHz and 64GB memory. There is one GPU node with 8 K20x GPUs. Beta will leverage some existing infrastructure (minimal software via modules, storage, etc).
The Beta cluster uses the Lmod modules system to provide access to centrally installed software. If you used a cluster at U-M previously, then you should review the documentation for the module system as we have changed the configuration to match that used at most national clusters and most other university clusters.
In particular, you should use the command module keyword
to look for a module and do not use module available
to search for software, as module available
will only show software for which all the dependencies (or prerequisites) are already loaded.
So, to search for the software package FFTW, use
$ module keyword fftw
That will show which versions are installed and provide a command to determine what is needed to load it.
Please see our page on using the Lmod modules system for more details and examples.
There are two main categories of software available on the system: software that is installed as part of the installation of the operating system and software that is installed separately. No special action is needed to use the software installed with the operating system. The separately installed software is set up so that you will use a module to use it. The module will set up the environment and make the software available. We do it this way to enable having multiple versions of the same package and to avoid having conflicts between software packages that have mutually exclusive system requirements.
Many of the software packages that are licensed for use on ARC clusters are licensed for a limited number of concurrent uses. If you will use one of those packages, then you must request a license or licenses in your submission script. As an example, to request one Stata license, you would use
#SBATCH --licenses=stata@slurmdb:1
The list of software can be found from Beta by using the command
$ scontrol show licenses
This section will be updated when system level changes are made to Beta. There are currently no updates.
Users must manage data appropriately in their various locations:
The Beta high-performance computing system at the University of Michigan has been built to provide a preview of the Great Lakes HPC cluster. Beta has the same security stance as the Flux cluster.
Applications and data are protected by secure physical facilities and infrastructure as well as a variety of network and security monitoring systems. These systems provide basic but important security measures including: