Job Overview
About WFU
WFU DEAC Cluster
Cluster Users
Winston-Salem NC
Apply for Job
|
Job Description
Wake Forest University seeks a systems analyst with experience in using
and supporting scientific computing on high performance computing clusters.
The successful candidate will not only participate in the operational
administration of our Linux cluster, but will also serve as a resource
for faculty and students learning to optimize their codes for parallel
computing architectures using both multithreaded and message passing
models. A Masters degree in Computer Science, Physics, Biophysics, or
other approved computational science is required, as is experience in
scientific and/or parallel programming and Linux system administration
equivalent to
SAGE Level 3
or better.
About Wake Forest University
Wake Forest University is a private university with about 4300
undergraduates and more than 700 graduate students, in addition to
professional schools of medicine, law, business, and divinity. Wake
Forest University ranks 30th among national universities-doctoral in
the new edition of US News & World Report's guide, "America's Best
Colleges." The annual guide gives Wake Forest high marks for its
small classes, low student-faculty ratio, high graduation and retention
rates, financial resources and alumni giving.
The university is a leader in the use of technology in higher education.
It was the first university to issues two laptop computers to each
undergraduate -- one when they enter the university, replaced by a
second at the beginning of the third year. Faculty receive a laptop
computer every second year (currently a Lenovo ThinkPad T60p, 2.2GHz
Core 2 Duo with 2Gb of RAM). Forbes.com reported that The Princeton
Review ranked Wake Forest University second in its October, 2003 report
on "America's Most Connected Campuses," a detailed survey of Internet
use in higher education.
About the WFU DEAC Cluster
The WFU DEAC HPC cluster is a research computing environment that is
centrally maintained by the University. The cluster administrator team
currently consists of Timothy Miller, who obtained his doctoral degree in
Physics using HPC clusters. This computational experience affords an
understanding of user operational needs and allows for a better and
more effective match of cluster design to the problems being studied.
WFU DEAC provides unique capabilities to campus researchers that are
not available from general campus computing: high speed networking
infrastructure, large scale storage and computational capacity.
Architecturally, WFU DEAC is a Linux-based Beowulf style cluster
consisting of 400 processors with 1 GB RAM per processor on 163 nodes
and 224 processors with 512MB RAM per processor on 28 nodes. Each node
has gigabit Ethernet connectivity. A subset of these nodes uses
specialized, high speed, low latency interconnects: 24 nodes (96 processors)
use the Infiniband based standard, 16 nodes (32 processors) use the
Myrinet based technology. Currently, all user accessible nodes (login
and computational) have direct access to 15 TB of usable storage
available through the high performance, parallel filesystem.
The WFU DEAC cluster design allows for a great deal of scalability in
the key areas of storage and computational nodes. We use IBM's General
Parallel File System (GPFS) software to manage and present the disk
storage connected to the cluster. GPFS provides every key function
that a cluster requires for its data access: fault tolerance, redundancy,
transparent maintenance, scalability in performance (multiple GB/s),
and scalability in capacity (2 PB tested limit).
As a whole, these technologies and resources allow WFU DEAC cluster
researchers to explore a great many computation problems that exist in
research today. The Myrinet and Infiniband technologies provide the
low latency, high bandwidth communication that is essential for difficult
parallel processing problems (fluid dynamics, thermodynamics, black
hole dynamics). The high performance and large scale filesystem provides
the backbone upon which users can tackle the large data set problems
(Monte Carlo simulations, genetics, bioinformatics, nuclear/particle
physics). Of course, the cluster is also well suited to ``traditional,
single processor problems that have no preconceived parallelization.
With multi-core processors and compiler technology, even traditional
software can see some performance gains using the WFU DEAC cluster and
the compilers we license through compiler vector optimizations.
About the Cluster Users
Currently, the WFU DEAC cluster supports users from the Physics, Computer Science,
Chemistry, Biology, Biomedical Engineering, and Statistical Genetics. Information
for these departments can be found at their respective websites:
About the Winston-Salem area
For more information about the campus and the city, please visit:
Apply for the Position
To apply for the position, please go to the WFU
Human Resources Online Employment System.
|
|
