The Clark Science Center’s Computer and Technical Services group (CATS) at Smith College maintains and supports a High Performance Computing (HPC) infrastructure. If you need to make use of HPC resources please contact us at firstname.lastname@example.org.
This HPC system currently consists three to four separate systems: an Apple Server cluster and a high-end Dell server running VMWare. The Apple Cluster consists of 12 servers each with eight 2.8 GHz processors; the head node has 14GB memory and the eleven worker nodes each have 6GB. The Dell server consists of 512GB of RAM and 36 processing cores. The virtual infrastructure on it allows CATS to make custom servers for multiple departments tailored to end-users specific computational needs. It also allows for robust, on-the-fly reallocation of unused resources to high-priority projects, and provides siginificant capacity for future expansion. Please read below for a description of other resources available.
Current HPC Hosts
Xgrid (aka mathgrid) Cluster [being decommissioned – being replace by newer Mac Cluster (see below)]
Consisting 12 Apple Xservers (model 2,1); 96 processors (268 GHz); 78GB RAM; running OS X 10.7.5; Xgrid.
Now only 2 Apple Xservers (model 2,1); 16 processors (22 GHz); 16GB RAM; running OS X 10.7.5; Xgrid.
Software being used:
ADF, Biopython, Gaussian, Matlab, Mathematica, NetLogo, Stata, R, RAxML, ORCA, R, & Stata
Virtual Host HPC Server
512GB RAM, 2×18 core processors (36 total) and 10TB of on-board storage. Virtualization allows the flexibility of running Windows or Linux OS’s at the same time. Resources can be split up and allocated to allow a wide range of server options.
Three experimental systems are being developed and tested. Specs so far…
3 Apple Xservers (model 3,1); 24 processors (x GHz); 18GB RAM; running OS X 10.11.6; Hadoop (vX.x)
Software being used: ADF, Matlab, Mathematica, NetLogo, R, Stata…
Netlogo server: 1 Dell server; (specs?) running Netlogo for Economics department
Hadoop, Apache Spark Cluster: 4 Dell servers; ~100 cores, 350GB RAM, Ubuntu 16.04 for spring 2017 CS course(s).
Cluster management tools: Xgrid, Hadoop, HTCondor, VMware, ganglia, PRTG
Biology — bioinformatics and genetic analyses using: Seqman NGen, Muscle, SOAPDenovo, SPAdes and other packages
Chemistry — calculational chemistry using: Gaussian, ORCA, ADF
Economics — Monte Carlo and other modeling simulations using: Matlab, Mathematica, NetLogo, Stata
Engineering — modeling using Ansys
Physics — Monte Carlo simulations using: Matlab
Statistics and Data Sciences — R
>GPU targeted HPC for image processing.
Potential primary users:
Spatial Analysis Lab — Running ArcGIS to analyze high resolution LiDAR data;
Biology/Neurology — Light Sheet Microscopy image processing, Data Sciences applications.
Economics — Planning to analyze large data sets using StataMP
Geosciences — possibly computing exoplanets orbital dynamics, also seismic-related calculations, groundwater modeling
>Increased use of cloud-base resources such as: AWS, Azure or Google
>Utilize resources at MGHPCC.