High Performance Computing
There are several high performance compute clusters available to UMass Boston researchers. These include two local clusters ('chimera', and 'gibbs') as well as a UMass Amherst managed shared cluster 'Unity' housed at the MGHPCC.
Chimera is a heterogeneous distributed memory compute cluster.
Gibbs is a heterogeneous distributed memory GPU-based high performance compute cluster.
Unity Unity is a large cluster shared with a number of other Universities and managed by the University of Massachusetts Amherst. It is located at the MGHPCC in Holyoke, MA.
Important note to users: In papers you publish using the High Performance Computing systems here, please acknowledge "the use of the supercomputing facilities managed by the Research Computing Department at the University of Massachusetts Boston." Also, please send the citation to email@example.com, so that they can be tracked and included in the bibliography.
All HPC users have access to limited storage within their 'home' directories, and each system has distinct home directories accessible only to the compute nodes of that cluster. In addition, there is a separate fileserver hpcstor1 which is accessible to all clusters and to all nodes within the clusters. This server has moderately better performance than the home directories, and any I/O intensive work should utilize this space (known as 'scratch' space) rather than home directories. In addition, there are a number of individually owned fileservers which are accessible from the compute clusters. If you have substantial space requirements, purchasing your own storage may be your best option. Please contact us for further information.
N.B. We do not provide backups of user data at this time. All users are responsible for maintaining their own backups. Scratch space is subject to purging as necessary.