HPC - Technology
High Performance Computing
The HPC Center at NPS strives to promote scientific computing at NPS by providing support to researchers and departments who wish to engage in scientific computing, and establish NPS as a nationally recognized HPC "Center of Excellence".
Supercomputing at NPS
NPS's supercomputer was installed on campus in 2009, and was named "Hamming" after the internationally renowned mathematician Richard Hamming, who was a Professor of Mathematics at NPS from 1976 until his death in 1998. The Hamming supercomputer is used by NPS faculty and students who require enormous amounts of computing power to solve a wide variety of problems related to thesis and research projects. Hamming contains 3,178 computing cores. A typical laptop computer today may contain four cores, so Hamming possesses the computing power of 800 laptops tied together to solve various types of problems. The presence of this computer and capability on campus has attracted several faculty members to NPS who require this type of higher-level equipment in order to perform their research and teaching duties.
The Hamming supercomputer is a "hybrid cluster", referring to the multiple nodes within Hamming possessing different hardware specifications as well as different capabilities and uses. Specifically, Hamming's capabilities can be divided into three different nodes/subsystems: MPI, Grace, and GPU.
Message Passing Interface (MPI) is a program model used in supercomputing that allows synchronization and communication between multiple processes during parallel computing. MPI provides high performance, scalability, and portability through Hamming’s cluster of thousands of core processors that can work together simultaneously to perform in-depth, complex calculations. This "number crunching" capability has been helpful in a number of different NPS applications, including global and regional weather modeling and simulations, an important research area for the DOD.
In the fall of 2015, HPC administrators installed the Apache Hadoop software framework into the Hamming supercomputer to develop a new subsystem. The new subsystem was named “Grace” after RDML Grace Hopper, an internationally renowned computer scientist and retired U.S. Navy veteran. Grace uses Hadoop’s distributed processing and data locality software to distribute very large data sets across multiple supercomputer nodes, allowing the data to be processed faster and more efficiently than most traditional supercomputer architectures. This "big data" capability allows the Hamming supercomputer to process and mine large amounts of data quickly, similar to commercial Internet search engines, and will an important tool for future NPS research. Access to the Grace nodes is currently restricted to a small group of NPS staff and researchers working on data analytics, but questions can be directed to the HPC administrators at firstname.lastname@example.org.
The Hamming supercomputer is also equipped with graphics processing unit (GPU) nodes, which provide visualization and graphical processing of research data sets. This capability is useful in providing a visual component to NPS research. In addition to graphics, Hamming’s GPU capability can also perform MPI parallel processing.
Academic Benefits of Hamming
The presence of this supercomputer on campus benefits students and faculty in many ways. One such benefit is the availability of high-powered computer software used in various NPS classes. Specialized software packages that can only be run on a supercomputer have been installed on Hamming to aid in instruction. Without these software packages, NPS faculty and students would be unable to understand and solve state-of-the-art problems relevant to DoD mission. An example of a software package is a commercial product named ANSYS CFX, which allows students to solve and understand problems related to fluid dynamics. For example, ANSYS CFX can simulate water flowing past ship hulls, a problem that is extremely relevant to the design of Navy vessels. Several classes at NPS rely on Hamming's capabilities to aid in graduate-level instruction every year. These classes include "Computer Architecture" in the Computer Science Department, "Fluid Dynamics" in the Mechanical Engineering Department, "Computer Systems" in the Electrical Engineering Department, and "Scientific Computing" in the Mathematics Department, among others.
NPS faculty members also benefit from using Hamming in their research projects. There are at least a couple dozen faculty research projects extensively using Hamming every fiscal year. Students additionally benefit from faculty-use of Hamming because their thesis projects often stem from their advisor's research project. Over the last few years, dozens of student theses have been produced using Hamming. A small sampling of thesis topics include weather forecasting, polar ice prediction, modeling of helicopter rotors, data mining (i.e., extraction of critical data from very large data sets), and the computation and solution of complicated mathematical equations. Without the Hamming supercomputer at NPS, all of these DOD-relevant research projects could not be completed.
For more information on how Hamming is used by NPS faculty and students, please visit the NPS Research Projects page.