Table of Contents
-
WELCOME:
Dennis
Aebersold,
University of Oklahoma
-
KEYNOTE:
José
Muñoz,
National Science Foundation
-
PLENARY:
Henry Neeman,
University of Oklahoma
-
PLENARY:
Michael Mascagni,
Florida State University
-
PLENARY:
Stephen Wheat,
Intel Corp
-
Joshua Alexander,
University of Oklahoma
-
John Antonio,
University of Oklahoma
-
Keith Brewster,
University of Oklahoma
-
Dana Brunson,
Oklahoma State University
-
Karen Camarda,
Washburn University
-
Wesley Emeneker,
University of Arkansas
-
Jeni Fan,
University of Oklahoma
-
Robert Ferdinand,
East Central University
-
Larry Fisher,
Creative Consultants
-
Dan Fraser,
University of Chicago
-
Roger Goff,
Sun Microsystems
-
Paul Gray,
University of Northern Iowa
-
Tim Handy,
University of Central Oklahoma
-
Takumi Hawa,
University of Oklahoma
-
Scott Lathrop,
TeraGrid
-
Evan Lemley,
University of Central Oklahoma
-
William Lu,
Platform Computing
-
Kyran (Kim) Mish,
University of Oklahoma
-
Greg Monaco,
Great Plains Network
-
Jeff Pummill,
University of Arkansas
-
Jeff Rufinus,
Widener University
-
Susan
J. Schroeder,
University of Oklahoma
-
Horst Severini,
University of Oklahoma
-
Dan Stanzione,
Arizona State University
-
Bradley C. Wallet,
University of Oklahoma
-
Dan Weber,
Tinker Air Force Base
-
Kenji Yoshigoe,
University of Arkansas at Little Rock
Other speakers to be announced
KEYNOTE
SPEAKER
Topic:
"High Performance Computing and
Cyberinfrastructure Activities at
the National Science Foundation"
Slides:
available after the Symposium
Talk Abstract
The
National
Science Foundation
(NSF)
has a long history of
supporting
High Performance Computing (HPC)
and making the technology available to
the open science and engineering communities.
The NSF
Cyberinfrastructure
Vision document
presents other CI components
that are meant to complement
the HPC investments
and
create an environment consistent with
the needs of the 21st century.
This presentation will discuss
where we are in the HPC area
as well as the other CI Vision areas,
in particular
the new activities in Data
as well as where more work is required
in order to achieve the CI Vision.
Biography
José
Muñoz
is Deputy Director of the
Office
of Cyberinfrastructure
(OCI)
at the
National
Science Foundation.
Prior to coming to NSF in February 2004,
Dr.
Muñoz
was Director of
Simulation and Computer Science
for the
Advanced Simulation and Computing
program at the
US
Department of Energy
(DOE)/National
Nuclear Security Administration
(NNSA),
and was at the
Defense
Advanced Research Projects Agency
(DARPA)
prior to
DOE.
Dr.
Muñoz
received his PhD in
Computer
Science
from the
University
of Connecticut
in 1984,
and his BSc in
Mechanical Engineering
from
New
York University
in 1967.
PLENARY
SPEAKERS
Director
OU
Supercomputing Center for Education
& Research (OSCER)
University
of Oklahoma
Topic:
"OSCER State of the Center Address"
Slides:
PowerPoint
PDF
Talk Abstract
The
OU
Supercomputing Center for
Education & Research
(OSCER)
celebrated its 7th anniversary
on August 31 2008.
In this report,
we examine
what OSCER is,
what OSCER does,
and where OSCER is going.
Biography
Dr.
Henry Neeman
is the
Director of the
OU
Supercomputing Center for Education &
Research
and
an adjunct assistant professor in the
School of
Computer Science
at the
University of
Oklahoma.
He received his BS in computer science
and his BA in statistics
with a minor in mathematics
from the
State
University of New York at Buffalo
in 1987,
his MS in CS from the
University of
Illinois at Urbana-Champaign
in 1990
and his PhD in CS from UIUC in 1996.
Prior to coming to OU,
Dr. Neeman was a postdoctoral research
associate at the
National
Center for Supercomputing Applications
at UIUC,
and before that served as
a graduate research assistant
both at NCSA
and at the
Center for
Supercomputing Research &
Development.
In addition to his own teaching and research,
Dr. Neeman collaborates with
dozens of research groups,
applying High Performance Computing techniques
in fields such as
numerical weather prediction,
bioinformatics and genomics,
data mining,
high energy physics,
astronomy,
nanotechnology,
petroleum reservoir management,
river basin modeling
and engineering optimization.
He serves as an ad hoc advisor
to student researchers
in many of these fields.
Dr. Neeman's research interests include
high performance computing,
scientific computing,
parallel and distributed computing,
structured adaptive mesh refinement
and
scientific visualization.
Professor
Department
of Computer Science
Florida State
University
Plenary Topic:
"Random Number Generation:
A Practitioner's Overview"
Slides:
PDF
Plenary Talk Abstract
We will look at random number generation
from the point-of-view of
Monte Carlo computations.
Thus, we will examine
several serial methods of
pseudorandom number generation
and two different parallelization techniques.
Among the techniques discussed with be
"parameterization,"
which forms the basis for
the Scalable Parallel Random Number Generators
(SPRNG)
library.
SPRNG was developed several years ago
by the author,
and has become widely used
within the international Monte Carlo community.
SPRNG is briefly described,
and the lecture ends with
a short revue of quasirandom number generation.
Quasirandom numbers offer
many Monte Carlo applications
the advantage of superior convergence rates.
Breakout Topic:
"Novel
Stochastic Methods in
Biochemical Electrostatics"
Slides:
PDF
Breakout Talk Abstract
We will present a Monte Carlo method for
solving boundary value problems (BVPs)
involving the Poisson-Boltzmann equation (PBE).
Such BVPs arise in many situations where
the calculation of electrostatic properties of
solvated large molecules.
The PBE is one of the implicit solvent models,
and has accurately modeled electrostatics over
a wide range of ionic solvent concentrations.
With the new method
we compare the algorithmic and computational
properties of this algorithm
to more commonly used
deterministic,
techniques,
and we present some computational results.
This work is part of an ongoing collaboration
with several
Florida
State University
faculty members,
students,
and collaborators at the
Russian
Academy of Sciences
and at the
University
of Toulon and Var.
Biography:
coming soon
Senior Director, High Performance Computing
Intel
Topic:
"Insatiable versus Possible:
Challenges on the Road to ExaScale"
Slides:
PDF
Talk Abstract
With the recent establishment of
computational science as
an equal third peer to
theory and experimentation
in the advancement of science and engineering,
the appetite for computational capability
has become insatiable.
There seems to be no end to
the rate of growth for the
computational capability of
the leading systems deployed annually
around the world.
Indeed,
maintaining parity with the competition,
let alone
establishing and retaining leadership,
has institutions continually on the move
regarding their own computational capacity.
In this talk,
we will address
how Moore's law plays into this phenomenon,
what other observations
we can make about the growth rate,
and
what barriers have been identified
regarding continued growth
over the next ten years.
We'll address the implications of
resolving these barriers.
And we'll close with
making the challenges and their resolutions
personal to the individuals in the audience.
Biography
Dr. Stephen Wheat is
the Senior Director of
Intel's
High Performance Computing
Platform Organization.
He is responsible for
the development of
Intel's HPC strategy
and the pursuit of that strategy
through
platform architecture,
software,
tools,
sales and marketing,
and
eco-system development and collaborations.
Dr. Wheat has a wide breadth of experience
that gives him a unique perspective
in understanding large scale HPC deployments.
He was the Advanced Development manager
for the Storage Components Division,
the manager of
the RAID Products Development group,
the manager of
the Workstation Products Group
software and validation groups,
and manager of
the systems software group
within
the Supercomputing Systems Division (SSD).
At SSD,
he was a Product Line Architect
and was
the systems software architect for
the ASCI Red system.
Before joining Intel in 1995,
Dr. Wheat worked at
Sandia
National Laboratories,
performing leading research in
distributed systems software,
where he created and led
the SUNMOS and PUMA/Cougar programs.
Dr. Wheat is a
Gordon
Bell Prize
winner
and has been awarded Intel's prestigious
Achievement Award.
He has a patent in Dynamic Load Balancing
in HPC systems.
Dr. Wheat holds a Ph.D. in Computer Science
and has several publications on
the subjects of
load balancing,
inter-process communication,
and
parallel I/O in large-scale HPC systems.
Outside of Intel,
he is a commercial multi-engine pilot
and
a certified multi-engine flight instructor.
BREAKOUT
SPEAKERS
HPC Application Software Specialist
OU Information
Technology
University
of Oklahoma
Topic:
"Implementing Linux-enabled Condor in
Multiple Windows PC Labs"
(with
Horst Severini)
Slides:
PowerPoint
PDF
Talk Abstract
At the
University
of Oklahoma
(OU),
Information
Technology
is completing a rollout of Condor,
a free opportunistic grid
middleware system,
across 775 desktop PCs in IT labs
all over campus.
OU's approach,
developed in cooperation with the
Research
Computing Facility
at the
University
of Nebraska Lincoln,
provides the full suite of Condor features,
including
automatic checkpointing,
suspension and migration as well as
I/O over the network
to disk on the originating machine.
These features are normally limited
to Unix/Linux installations,
but OU's approach allows them on
PCs running Windows as
the native operating system,
by leveraging
coLinux
as a mechanism for providing Linux as
a virtualized background service.
With these desktop PCs otherwise idle
approximately 80% of the time,
the Condor deployment is allowing OU
to get 5 times as much value
out of its desktop hardware.
Biography
Joshua Alexander is a
Computer Engineering
undergraduate at the
University
of Oklahoma.
He currently works with the Customer Services
division of
OU
Information Technology,
and also serves as
an undergraduate researcher
for the
OU
Supercomputing Center for Education &
Research (OSCER).
His current project for OSCER
involves both the OU IT
Condor
pool
and development of software tools
for deploying Condor at other institutions.
Professor
School
of Computer Science
University
of Oklahoma
Topic:
"Reconfigurable Versus Fixed Versus
Hybrid Architectures"
Slides:
PowerPoint
PDF
Talk Abstract
Until recently,
the use of reconfigurable computing
has been limited primarily to
embedded
High Performance Computing (HPC) applications,
for example,
signal processing applications
having
extremely high-throughput data streams
and
intensive computational requirements.
However,
reconfigurable computing technology
is now finding its way into
systems used to support
broader applications in the realm of HPC.
This talk will include
a brief overview of reconfigurable computing
and also provide rationale for why
reconfigurable computing is becoming
more universally viable.
An overview of the tools and techniques
required to harness
the full potential of
reconfigurable computing resources
will be provided.
The talk will then focus on
architectures that make use of
both fixed and reconfigurable
computational resources.
An example hybrid multi-core architecture
will be described
that can be configured
to optimally support
a wide variety of computational requirements,
ranging from
independent threads
to
massively parallel applications
requiring
intensive inter-processor communications.
Biography
John
K. Antonio
is Professor of
Computer
Science
at the
University
of Oklahoma.
He received his
BS, MS, and PhD degrees
in
Electrical
Engineering
from
Texas
A&M University.
From 1999 to 2006,
he served as
Director and Professor of
Computer Science
at
OU,
and from 2006 to 2008
he was Director of the
Institute for Oklahoma Technology Applications
(IOTA)
at OU.
Before joining OU,
he was on the faculty of
Electrical
and Computer Engineering
at
Purdue
University,
and he was also on the faculty of
Computer
Science
at
Texas
Tech University.
Dr. Antonio is a senior member of the
Institute
of Electrical and Electronics Engineers
(IEEE),
a member of the
Association
of Computing Machinery,
and an elected member of the
European
Academy of Sciences.
He is Associate Editor of the journal
IEEE
Transactions on Computers.
His academic research interests include:
embedded high performance computing;
low-power and power-aware computing;
reconfigurable computing;
parallel and distributed computing;
and
cluster computing.
Dr. Antonio has co-authored
over 90 publications and reports
in the above and related areas.
Numerous agencies and companies
have supported his research over the years.
He has been PI or Co-PI on
more than twenty sponsored research projects
totaling more that $2M.
In his role as Director of IOTA,
he managed and helped grow
a diverse research portfolio
having
over $7M in expenditures
and
over $20M in-force in annual funding.
Senior Research Scientist
Center for
Analysis & Prediction of Storms
University of
Oklahoma
Topic:
"Using the LEAD Portal for
Customized Weather Forecasts
on the TeraGrid"
Slides:
PowerPoint
PDF
Talk Abstract
The
Linked
Environments for Atmospheric Discovery
(LEAD)
portal is
a web tool for
accessing and visualizing weather data,
providing instruction in meteorology,
and
conducting
numerical weather forecasting experiments.
As part of the
Storm
Prediction Center's
annual Spring Program,
the
Center
for Analysis and Prediction of Storms
(CAPS)
used the LEAD portal
to generate
customized high resolution
numerical weather forecasts
that focused on the region
under threat for that day.
The forecasts ran
using the
BigRed
TeraGrid
machine at the
Indiana
University,
and the interaction
with the
TeraGrid
resources was handled seamlessly by
the LEAD
Portal.
Post-processing was done on
TopDawg
of
OU
Supercomputing Center for
Education & Research
(OSCER)
and results were available
in real-time for
the experiment forecasters to discuss
in the daily weather briefing.
Biography
Keith Brewster is a Senior Research Scientist
at the
Center
for Analysis and Prediction of Storms
at the
University
of Oklahoma
and an Adjunct Associate
Professor in the
OU School of
Meteorology.
His research involves
data assimilation of
advanced observing systems
for high resolution
numerical weather analysis and prediction,
including data from
Doppler
radars,
satellites,
wind profilers,
aircraft
and
surface mesonet systems.
He earned an M.S. and Ph.D. in Meteorology
from the
University
of Oklahoma
and a B.S. from the
University
of Utah.
Senior Systems Engineer
High
Performance Computing Center
Oklahoma
State University
Topic:
"Birds of a Feather Session:
So You Want to Deploy
a Production Cluster"
(with
Jeff Pummill)
Slides:
PDF
BoF Abstract:
This BoF is intended as an introduction to
the many components that make up
a contemporary cluster environment.
The presentation and accompanying discussion
will address topics such as:
how to choose hardware type(s),
the various software stacks available,
pros and cons of
the various applications
that are used on clusters,
administrative tips and tricks,
user support advice,
and hopefully a lively debate at the end.
This BoF is not intended to define
what should and should not be deployed;
rather,
we will present
the many factors and considerations
involved in
deploying a successful cluster,
and we will outline
the various rewards and pitfalls
along the way.
Biography
Dana Brunson
oversees the
High
Performance Computing Center
at
Oklahoma
State University
Before transitioning to
High Performance Computing
in the fall of 2007,
she taught mathematics
and served as systems administrator
for the
Mathematics department
at
OSU.
She earned her Ph.D.
in Numerical Analysis
at the
University
of Texas at Austin
in 2005 and her M.S. and B.S. in Mathematics
from
Oklahoma
State University.
Associate Professor
Department
of Physics & Astronomy
Washburn
University
Topic:
"Supercomputing at a
Non-PhD Granting Institution"
Slides:
PDF
Talk Abstract:
With funds from an internal "Innovation Grant,"
faculty members at
Washburn
University
are developing a
High
Performance Academic Computing Environment
(HiPACE).
HiPACE was proposed to enrich and enliven
the science and technology education of
all Washburn students,
to support the scholarly activities of
Washburn faculty, students, and staff,
and to open outreach opportunities
by making advanced technology training
available to K-12 educators.
In this talk,
I will discuss the progress being made
in fulfilling the goals of HiPACE,
as well as some difficulties that have arisen.
Biography
Dr.
Karen Camarda
is an associate professor in the
Department
of Physics & Astronomy
at
Washburn
University.
She received her BS in physics from the
University
of California San Diego
in 1991,
her MS in physics from the
University
of Illinois at Urbana-Champaign
in 1992,
and her PhD in physics from UIUC in 1998.
Dr. Camarda serves as
the chair of the steering committee for
Washburn University's
High
Performance Academic Computing Environment
(HiPACE).
Her research interests lie in
the field of numerical relativity.
Graduate Research Assistant
Department
of Computer Science &
Computer Engineering
University
of Arkansas
Topic:
"Cluster Scheduling:
Making Everybody Happy All the Time
(Yeah Right!)"
Slides:
available after the Symposium
Talk Abstract
Scheduling cluster jobs
is more art than science.
Policies designed to run
as many jobs as possible
negatively impact large jobs.
Policies designed to accommodate large jobs
work at the expense of decreased utilization.
Furthermore,
users figure out how to work the system
to make their jobs run as soon as possible.
In this talk,
we look at a few scheduling policies,
and see how they affect both users and jobs.
Additionally,
we look at some research being done at the
University
of Arkansas
to predict future workloads
based on current data.
Biography
Wesley Emeneker
is a Ph.D. student at the
University
of Arkansas Fayetteville.
He is the principal architect of
Dynamic
Virtual Clustering,
a system that uses virtual machines
to run cluster jobs.
His current research is looking at
describing exactly how virtual machines
impact application performance,
and how this knowledge
can be applied to batch scheduling
to improve the cluster "experience"
for users and jobs.
Graduate Research Assistant
Department
of Psychology
University
of Oklahoma
Topic:
"Integrating Bayesian Inference with Semantics"
!<---
Slides:
available after the Symposium
PDF
--->
Talk Abstract
HyGene
(Thomas, Dougherty, Sprenger, & Harbison,
2008)
is a theory of
hypothesis generation, evaluation and testing
that bridges the gap between
traditional theory and research in
judgment and decision making
with research in
ecological and cognitive psychology.
This presentation will briefly review HyGene,
but will primarily focus on
future research directions
that require supercomputing.
Specifically,
we will discuss a project aimed at
developing a model of
hypothesis generation and judgment
that merges
our Bayesian-like hypothesis generation model
(HyGene)
with
state-of-the-art high-dimensional models of
semantic knowledge
(e.g., Landauer & Dumais, 1997).
Our goal is to develop
a model of hypothesis generation and judgment
that capitalizes on
the power of Bayesian reasoning
while exploiting the semantics inherent in
natural language.
Biography
Born in Beijing, China,
Jeni Fan
moved to the Washington DC metro area
at age 9.
She
earned her BS in Psychology at the
University
of Maryland
in 2005
and her MS in Cognitive Psychology at the
University
of Oklahoma
in 2008.
Between these degrees,
she worked for
Bloomberg
LP
in New York City.
Her areas of interest are
Decision Theory and Behavioral Economics,
focusing on
(1) investigating the effects of context on
decision making and option evaluation;
(2) assessing the role of utility in
hypothesis generation, evaluation, and search,
as well as choice behavior;
(3) working with the
Department of Economics
(Game Theory specifically)
to establish an integration between
classic economic theories
and
more behavioral decision making phenomena.
Associate Professor
Department
of Mathematics
East
Central University
Topic:
"Finite Element Solution of
a Groundwater Contaminant Model"
Slides:
PDF
Talk Abstract
The model presented takes the form of
a coupled system of two nonlinear
partial differential equations
describing the dynamics of
contaminant in groundwater
flowing through fissures (cracks) in
a rock matrix,
leading to
the contaminant traveling and diffusing
along the length of the fissure
and also into the surrounding rock matrix.
This diffused contaminant can
reach a water body,
for example,
that is being used as
a drinking water source for
local area residents and/or livestock,
which can thereby cause a health hazard.
A new feature of this model is
an added dimension in
the rock matrix diffusion term.
A Galerkin finite method using a triangulation
is used to approximate
the model solution in L-2 norm.
In the future,
it is hoped that this scheme can be used
to estimate model parameters
using an inverse method procedure.
Both solution and parameter approximation
will use large amounts of computation.
Biography
Robert Ferdinand obtained his PhD
in Applied Mathematics
from the
University
of Louisiana
in 1999.
His areas of interest include
mathematical modeling of
physical and biological processes,
in which
numerical schemes are
used to computationally approximate
model solutions:
for example,
the inverse method is applied to
numerically estimate model parameters,
which
involves substantial computing.
His theoretical work involves
perturbation techniques to investigate
long-term behavior of model solutions.
Owner
Creative Consultants
Topic:
"Careers in a Creative Destruction World"
Slides:
PowerPoint
PDF
Talk Abstract
Creative destruction occurs when
a product of less quality and less cost replaces
a product of higher quality
due to convenience and/or attractiveness.
Existing companies
fall victim to creative destruction
because they are
too busy maintaining the status quo;
they never see it coming.
Several examples and stories of
the impact of creative destruction
will be provided,
including potential
creative destruction products
such as
buckyballs,
carbon nanofibers,
HHO gas,
and
UAVs.
You will be encouraged to identify
creative destruction events in
the computer and software industry.
Biography
Larry Fisher
is a retired state employee
with over 30 years of
management development and training experience.
He has designed and taught courses nationally
for the
US Air Force;
the
US Postal Service;
the
Domestic Policy Association;
the
University
of Oklahoma;
Oklahoma
State University;
Wichita
State University;
Oklahoma
Gas and Electric,
the
Municipal
Electric Systems of Oklahoma;
the states of
Oklahoma,
Kansas,
South Dakota,
Texas,
and
Ohio;
and the
Kiowa,
Seminole,
and Cheyenne/Arapaho
Indian Nations.
He administered
statewide management training
for the state of Oklahoma.
For 16 years,
he worked in a variety of
professional positions for
the
University
of Oklahoma.
He is known nationally through memberships
and as an officer in the
American
Society for Training and Development,
the
National
Association for
Government Training and Development,
and the
International
Association for
Continuing Education and Training.
He currently teaches
management and related topics
to a variety of clients
including
the
Keller
Graduate School of Management
at
DeVry
University,
Municipal
Electric Systems of Oklahoma,
Tinker
Air Force Base,
Rose
State College,
and many others.
He has a BS degree in Chemistry from
Oklahoma
State University
and a Masters in Public Administration from the
University
of Oklahoma.
Senior Fellow
Computation
Institute
University
of Chicago
Topic:
"What Happens When Cloud Computing
Meets High Performance Computing"
Slides:
PowerPoint
PDF
Talk Abstract
The computing community
is still struggling to comprehend
not only how to fully realize
the promise of cloud computing,
but what that promise actually is.
Even less understood
is how cloud technology
might work together with
high performance computing (HPC).
To explore these issues,
the relationship between
HPC computing and Cloud computing
is considered,
along with
the value propositions provided by each.
Insights from this analysis
are then used to examine
several possible HPC-Cloud scenarios.
The
Globus
Toolkit
is relevant to these discussions
and will be introduced
both as an enabler for
Grid Technology
and also as
a provider of an open-source
Amazon
Elastic
Compute Cloud-like
capability.
Biography
Dan Fraser is a Senior Fellow at the
Computation
Institute
at the
University
of Chicago.
Currently he is PI of the
"Real Time Analysis of
Advanced Photon Source Data" project
and is also Director of the
Community
Driven Improvement of Globus Software
program for the
National
Science Foundation.
Formerly he was the
Senior Architect for Grid Middleware at
Sun
Microsystems
and the creator of Sun's
Technical
Computing Portal.
He has a PhD in Physics from
Utah
State University
and over a decade of experience working with
high performance science and commercial
applications.
HPC Architect
Sun
Microsystems
Topic:
"Managing Mountains of Data in
Large Scale HPC Systems"
Slides:
PDF
Talk Abstract
One of the biggest challenges
facing designers and users of HPC
solutions today is managing the
flow of the ever-increasing
amount of data being processed.
While parallel filesystems are
maturing and are being used more
broadly,
they only solve part of the problem.
A more holistic approach
&mdash
one that encompasses not
only the needs for fast scratch
space but the requirements for
archival, visualization, and end
users as well
&mdash
is required.
A proven solution that fits today's
environments will be presented,
along with a look at future
directions for HPC data
management technologies.
Biography
Roger Goff
(roger.goff@sun.com)
is a HPC Architect in the Systems
Practice at Sun Microsystems.
Roger has been in the high
performance technical computing
business since 1987.
His current
interests include high
performance computing clusters
and storage solutions for large
scale HPC solutions.
Roger
designed numerous systems that
debuted in the top 100 systems of
the
Top 500
fastest
supercomputers list,
including four
systems that debuted in the top
twelve and two in the top five.
Roger has presented multiple
times at the Linux World
Conference and Expo, HP Users
Group, Dell Technology Summits
and Sun Customer Engineer
Conferences.
Roger has M.S. and
B.S. degrees in Computer Science
from
Virginia Tech.
Associate Professor
Department of
Computer Science
University of
Northern Iowa
Topic:
"High Performance Computing in
the Core Computer Science Curriculum"
Slides:
PowerPoint
PDF
Talk Abstract
Hardware chipset vendors
aren't making single-core chips any more
for end user systems,
so why are we still focusing on
single-core architectures
in our computer science curriculum?
High Performance Computing (HPC)
is moving from multi-core to many-core;
from traditional architectures to
specialized processors such as GPUs.
This talk will discuss efforts
actively focusing on
infusing our undergraduate curriculum
in a way that is commensurate with
the momentum of the HPC community.
We will also take a brief tour of
the tools and software
that are needed to prepare
the next-generation of researchers.
Biography
Paul
Gray
is an Associate Professor of
Computer
Science
at the
University
of Northern Iowa.
He is the chair of the
SC
(SuperComputing) Conference
Education Program
and instructs summer workshops on
parallel computing education with the
Supercomputing Education program
efforts.
His current efforts combine the
Open
Science Grid
and
TeraGrid
with educational endeavors that revolve around
LittleFe
bringing aspects of grid computing into
the high school and undergraduate curriculum.
Undergraduate Student
Department
of Engineering & Physics
University of
Central Oklahoma
Topic:
"Computational Aspects of
Modeling Fluid Flow in Micro-junctions"
Slides:
PDF
Talk Abstract
This talk will focus on
the computational efforts
that have been undertaken
by a research group at the
University
of Central Oklahoma,
along with
collaborators at the
University
of Oklahoma,
to model and simulate laminar flow
in microtubes and junctions.
In particular,
the set of programs and
scripts developed to automate
various processes involved in
the large number of
computational fluid dynamic (CFD)
runs required in this research project
will be described.
This discussion will allow
other researchers facing
similar problems
some solution techniques
in the area of automation of
geometry production and CFD related issues.
Biography
Tim Handy
is a senior engineering physics major at the
University of
Central Oklahoma.
His current research interests include
loss coefficients for
laminar flow in microbifurcations
and
flow through porous media.
His current plan is
to attend graduate school in
mechanical engineering
or
computational mathematics.
Assistant Professor
School
of Aerospace &
Mechanical Engineering
University of
Oklahoma
Topic:
"Nanoparticle Synthesis and
Assembly from
Atomistic Simulation Studies"
Slides:
PDF
Talk Abstract
Some 75% of chemical manufacturing processes
involve fine particles at some point.
Proper design and handling of
these fine particles
often makes the difference
between success and failure.
Careful attention to particle characteristics
during the design and operation of a facility
can significantly improve
environmental performance
and
increase profitability
by improving product yield and reducing waste.
Fabrication of the desired size
with a narrow size distribution,
and desired structure,
is seen as
one of the major challenges in
robust implementation of nanoscience
to a nanotechnology.
The two most obvious ways to control
the size of primary particles
grown from the vapor
are
either
to change the characteristic collision time
by dilution
or
to change the sintering time
by changing particle temperature.
In this talk,
sintering of nanoparticle aggregates
with the fractal dimension of
1 (wire),
1.9 (complex),
and
3 (compact)
is investigated using
molecular dynamics simulations.
The sintering times
normalized by the primary particle diameter
show a universal relationship
that only depends on
the number of particles in an aggregate
and
its fractal dimension.
This result is found to be consistent with
a continuum viscous flow mathematical model
that we developed.
The results for the sintering of
arbitrary fractal aggregates
can be approximated with
a power law modification of
the Frenkel viscous flow equation,
to include a dependence on
the number of particles in
a fractal aggregate and fractal dimension.
The role of surface passivation on
the rate of nanoparticle sintering
is also considered.
The presence of hydrogen on
the surface of a particle
significantly reduces surface tension.
In general,
the entire sintering time of coated particles
is about 3 to 5 times that of bare particles,
and
the viscous flow model describes
the dynamics of sintering of coated particles.
Also,
the approach of hydrogen coating is applied
to control the shape of the nanoparticle.
Finally,
electrostatically directed
nanoparticle assembly on
a field-generating substrate is studied.
Brownian motion and fluid convection
of nanoparticles,
as well as the interactions between
the charged nanoparticles
and
the patterned substrate,
including electrostatic force,
image force and van der Waals force,
are accounted for in the simulation.
Coverage selectivity is most sensitive to
electric field,
which is controlled by
the applied reverse bias voltage
across the p-n junction.
Biography
Takumi
Hawa
received his B.S., M.S., and Ph.D.
in Aeronautical Engineering
in 1994, 1997, and 1999 from
Rensselaer
Polytechnic Institute.
During a postdoctoral fellowship at the
Institute
for Mathematics and Its Applications
at the
University
of Minnesota
between 1999 and 2001,
he studied microfluidics and detonation.
He joined the
Center
for NanoEnergetics Research,
also
at the
University of Minnesota,
developed in 2001,
and researched
hydrogen surface passivated
silicon nanoparticles
to control the size of the nanoparticles
using Molecuar Dynamics Simulation.
In 2003,
he moved to the
University
of Maryland
and the
National
Institute of Standards & Technology
as a guest researcher
and has been studying
various passivation surfaces for nanoparticles,
such as solid coating and SAM (polymer).
In fall 2008,
He became an Assistant Professor in the
School
of Aerospace &
Mechanical Engineering
at the
University of
Oklahoma,
and currently studies
energetic materials,
AFM tip interaction with a substrate,
and
the phase stability of
hydrogen-coated silicon nanoparticles
as a function of size and shape.
Blue
Waters
Technical Program Manager for Education
Area Director for
Education,
Outreach & Training
TeraGrid
Topic:
"HPC University"
Slides:
PowerPoint
PDF
Talk Abstract
HPC University
(HPCU)
is a virtual organization focused on
high-quality, high-performance computing (HPC)
learning and workforce development
activities and resources.
HPC University
is designed to address the needs of
a large and diverse community
that includes:
K-20 educators and students;
undergraduate faculty and students;
graduate, post-doc and senior researchers;
administrators;
and
practitioners in all fields of study
related to HPC.
The content ranges from
introductory computational science
tools and resources,
to petascale level performance of
scientific research codes.
During the Symposium,
there will be a discussion of
the HPCU requirements
analysis, implementation, and
dissemination plans.
There will be a question and answer period
to solicit additional community input
and foster
increased collaboration and participation
within the community.
We invite all interested organizations
to join in developing effective strategies for
expanding and scaling-up the opportunities
to best serve
the computational science and HPC needs of
research and education communities.
Biography
Scott Lathrop
splits his time between being the
TeraGrid
Director of
Education,
Outreach & Training
(EOT)
at the
University
of Chicago/Argonne
National Laboratory,
and being the
Blue
Waters
Technical Program Manager for Education
for the
National
Center for Supercomputing Applications
(NCSA).
Lathrop has been involved in
high performance computing and
communications
activities
since 1986.
Lathrop coordinates
education, outreach and training activities
among the eleven Resource Providers
involved in the
TeraGrid
project.
He coordinates
undergraduate and graduate education activities
for the Blue Waters project.
Lathrop is Co-PI on the
National
Science Foundation
(NSF)
funded
Computational
Science Education Reference Desk
(CSERD),
a
Pathways
project of the
National
Science Digital Library
(NSDL)
program.
Lathrop coordinated the creation of
the
SC07-10
Education Program
through the
SC
Conference
(Supercomputing 20XX)
to assist
undergraduate faculty and high school teachers
with integrating
computational science
resources, tools, and methods
into the curriculum.
Professor
Department
of Engineering & Physics
University of
Central Oklahoma
Topic:
"Computational Aspects of
Modeling Fluid Flow in Micro-junctions"
Slides:
PDF
Talk Abstract
This talk will focus on
the computational efforts
that have been undertaken
by a research group at the
University
of Central Oklahoma,
along with
collaborators at the
University
of Oklahoma,
to model and simulate laminar flow
in microtubes and junctions.
In particular,
the set of programs and
scripts developed to automate
various processes involved in
the large number of
computational fluid dynamic (CFD)
runs required in this research project
will be described.
This discussion will allow
other researchers facing
similar problems
some solution techniques
in the area of automation of
geometry production and CFD related issues.
Biography
Evan Lemley
received his BA in Physics from
Hendrix
College
and MS and Ph.D
in Engineering (Mechanical) from the
University
of Arkansas.
His thesis
work was focused on modeling and simulation of
various neutron detectors.
Post graduation Evan worked for
the engineering consulting firm
Black &
Veatch
in a group responsible for
modeling coal power plants with
custom written software.
In August 1998,
Evan became an Assistant Professor in the
Department
of Engineering and Physics
(formerly Physics)
at the
University
of Central Oklahoma,
and has been there since,
teaching
mechanical engineering,
physics,
and
engineering computation
courses.
Early research at UCO was focused on
neutron transport in materials.
More recently,
Evan has been involved in simulation of
flow in microtubes and microjunctions
and
simulation of flow in porous networks.
Director, Industry Marketing
Platform
Computing
Topic:
"Using and Managing HPC Systems"
Slides:
PDF
Talk Abstract
Commodity hardware has become
the affordable building block for HPC systems.
Meanwhile,
scalable systems are complex
to use and to manage.
Platform Computing leverages
its 16 years of expertise to deliver
an end-to-end, affordable
HPC management solution
to truly unleash the power of HPC.
Click
here
for more information.
Biography
As marketing director at Platform Computing,
William Lu is focused on
government, research, education,
and the electronics industry.
Due to his deep technical background
and strong HPC experience,
he also leads a team of architects
at Platform to develop technical solutions.
During his 13-year tenure
at Platform Computing,
William has worked in product development,
professional services,
systems engineering,
and marketing.
William also has four years of HPC experience
at CERN and the University of Texas.
William has a Ph.D. in high energy physics.
Director
Fears
Structural Engineering Laboratory
Presidential Professor of
Structural Engineering
School
of Civil Engineering &
Environmental Science
University
of Oklahoma
Topic:
"Virtual Canaries in Virtual Coalmines:
Detecting Infrastructure Damage
via Computational Engineering"
Slides:
available after the Symposium
Talk Abstract
Monitoring the health of
our shared transportation infrastructure
is one of the most important research topics
in civil engineering today,
but this problem is complicated by
the fact that
real-world damage can quickly lead to
collapse of infrastructure,
with attendant loss of life and property.
The
Infrastructure Institute
at the
University
of Oklahoma
is currently using
advanced simulation techniques from
the field of computational engineering
to design and construct
custom sensor technologies
specifically tailored to detect
various forms of structural damage
before they can cause serious problems.
Examples applications using
fatigue and other damage models
will be presented,
along with a general overview of
the underlying technical problem
and
the attendant high-performance computing
technologies utilized to solve it.
Biography:
coming soon
Executive Director
Great
Plains Network
Topic:
"Roundtable:
Great Plains Network Regional CI Planning
(Initial Meeting)"
Slides:
available after the Symposium
Roundtable Abstract
The
Great
Plains Network
membership started with universities in
7 midwestern states,
and has
expanded to include
universities in Iowa and Minnesota,
as well as the state
research and education network of Wisconsin.
There is great potential
to share resources and
to collaborate across this wide area
(see
here,
for example),
and this roundtable will be used
to come up with preliminary
recommendations
the region discuss
potential new areas for future
regional collaboration,
pooling of resources to collaborate
as well as to pool and share resources.
Biography
Dr.
Greg Monaco
has held several positions with the
Great
Plains Network
since August 2000, when he joined GPN.
He began as Research Collaboration Coordinator,
and then was promoted to
Director for Research and Education.
Greg is currently the
Executive Director of GPN.
His resume can be found
here.
Senior HPC Administrator
Arkansas High
Performance Computing Center
University
of Arkansas
Topic:
"Birds of a Feather Session:
So You Want to Deploy
a Production Cluster"
(with
Dana Brunson)
Slides:
available after the Symposium
BoF Abstract:
This BoF is intended as an introduction to
the many components that make up
a contemporary cluster environment.
The presentation and accompanying discussion
will address topics such as:
how to choose hardware type(s),
the various software stacks available,
pros and cons of
the various applications
that are used on clusters,
administrative tips and tricks,
user support advice,
and hopefully a lively debate at the end.
This BoF is not intended to define
what should and should not be deployed;
rather,
we will present
the many factors and considerations
involved in
deploying a successful cluster,
and we will outline
the various rewards and pitfalls
along the way.
Biography
Jeff
Pummill
is the Senior HPC Administrator for the
Arkansas
High Performance Computing Center
at the
University
of Arkansas.
Prior to his position at the
UofA,
he spent 13 years in
the fields of
mechanical design and
structural analysis,
while also maintaining
a large number of Unix workstations
and a small Linux clusters
used for Finite Element Analysis.
His current areas of interest include
hardware architectures,
resource managers,
compilers,
and benchmarking tools.
He is also the
TeraGrid
Campus
Champion
for the
University
of Arkansas.
Associate Professor
Sciences
Division
Widener
University
Topic:
"ABINIT:
An Open Source Code for
Materials Scientists,
Computational Chemists,
and Solid State Physicists"
Slides:
available after the Symposium
Talk Abstract
ABINIT
is an open-source code
that can be
used to do
ab-initio calculations
on materials ranging from
atoms
to
clusters
to
periodic structures.
In this talk,
the ABINIT code will be introduced.
Examples will also be given.
Biography
Jeff Rufinus
obtained his Ph.D. in physics from the
University
of Wisconsin-Madison.
Jeff has been teaching computer science
and computational science at
Widener University
for the
past 8 years.
His research interests include magnetic
properties of diluted semiconductors,
spintronics
and
nanotechnology.
Assistant Professor
Department
of Chemistry & Biochemistry
University
of Oklahoma
Topic:
"Progress Towards
Predicting Viral RNA Structure from Sequence:
How Parallel Computing can Help Solve
the RNA Folding Problem"
Slides:
PowerPoint2007
PowerPoint2003
PDF
Talk Abstract
As genome sequencing projects produce
increasingly vast amounts of data,
the need for tools to interpret
genomic sequence information
at a structural level
becomes increasingly urgent.
Ribonucleic acid (RNA)
plays important roles in the
processing,
regulation,
and
transformation
of genetic information in cells.
RNA folds into
three-dimensional structures
and thus achieves
specificity in
molecular recognition and enzymatic activity.
Many viruses have RNA genomes,
such as HIV,
flu,
and hepatitis.
Satellite tobacco mosaic virus
is a small plant virus
and a good starting point for
improving predictions of viral RNA.
Viral RNA changes conformation during
replication,
translation,
and
encapsidation.
A single minimum free energy structure
is inadequate to describe
dynamic viral RNA structures.
Incorporating experimental restraints
from crystallography and chemical modification
can improve prediction of
different functional viral RNA structures.
The Wuchty algorithm
uses free energy minimization to calculate
all possible structures for
a given RNA sequence within
a narrow energy increment.
Modifications to the Wuchty algorithm include
nucleotide-specific restraints from
chemical modification experiments,
global restraints from crystallographic data,
and
parallelization of the computation.
These modifications will enable
wider exploration of the RNA folding landscape.
More information about viral RNA structures
will enable
rational design of
small molecules
or
RNA interference strategies
to inhibit
viral genome encapsidation and propagation.
Biography
Susan
Schroeder
began exploring
RNA energetics,
structure,
and
function
as an undergraduate in
Dr. Douglas Turner's
lab
at the
University
of Rochester
in Rochester NY.
As a graduate student in Dr. Turner's lab,
she continued to study
the thermodynamic stability of
RNA internal loops
and the structure of these loops by
nuclear magnetic resonance spectroscopy (NMR).
As an NIH postdoctoral fellow in
Dr. Peter Moore's
lab at
Yale
University,
she learned x-ray crystallography
and
molecular biology techniques
while probing the structures of
antibiotic resistance mutations in ribosomes
and
discovering new RNA binding sites of
novel drugs that target ribosomes.
As an assistant professor in
the
Department
of Chemistry and Biochemistry
at the
University
of Oklahoma
she is now applying her diverse skills to
the study of
viral RNA
structure,
function,
and energetics.
She gratefully acknowledges
research financial support from
the following agencies:
Pharmaceutical Research and Manufacturers of
America Foundation,
Oklahoma Center for the Advancement of
Science and Technology
Plant Science Research Program,
Oklahoma University Health Science Center
Institutional Research Grant from
the American Cancer Society,
and the
Department
of Chemistry and Biochemistry
at the
University
of Oklahoma.
Research Scientist
Department
of Physics & Astronomy
University
of Oklahoma
Topic:
"Implementing Linux-enabled Condor in
Multiple Windows PC Labs"
(with
Joshua Alexander)
Slides:
PowerPoint
PDF
Talk Abstract
At the
University
of Oklahoma
(OU),
Information
Technology
is completing a rollout of Condor,
a free opportunistic grid
middleware system,
across 775 desktop PCs in IT labs
all over campus.
OU's approach,
developed in cooperation with the
Research
Computing Facility
at the
University
of Nebraska Lincoln,
provides the full suite of Condor features,
including
automatic checkpointing,
suspension and migration as well as
I/O over the network
to disk on the originating machine.
These features are normally limited
to Unix/Linux installations,
but OU's approach allows them on
PCs running Windows as
the native operating system,
by leveraging
coLinux
as a mechanism for providing Linux as
a virtualized background service.
With these desktop PCs otherwise idle
approximately 80% of the time,
the Condor deployment is allowing OU
to get 5 times as much value
out of its desktop hardware.
Biography
Horst
Severini
got his Vordiplom (BS equivalent) in Physics at
the
University
of Wuerzburg
in Germany in 1988,
then went on to earn a Master of Science
in Physics in 1990
and
a Ph.D. in Particle Physics in 1997,
both at the
State
University of New York at Albany.
He is currently a Research Scientist in
the
High
Energy Physics group
at the
University
of Oklahoma,
and also the Grid Computing Coordinator at the
Oklahoma
Center for High Energy Physics
(OCHEP),
and the Associate Director for
Remote and Heterogeneous Computing at
OU
Supercomputing Center for
Education & Research
(OSCER).
Director
High
Performance Computing Initiative
Arizona
State University
Topic:
"A Scalable Framework for
Offline Parallel Debugging"
Slides:
PDF
Talk Abstract
As clusters get larger,
we have increasingly easy access
to run jobs on
thousands,
tens of thousands,
or even hundreds of thousands of cores.
However,
our ability to debug these jobs at scale
has not kept up with
the growth in hardware.
In this talk,
the
GDBase
framework for offline
debugging will be presented.
GDBase solves three problems in large scale
debugging:
(1) it integrates with batch systems
to allow debugging jobs to be run
without the need to interrupt
production operation;
(2) it moves debugging from online to offline,
to reduce the amount of system time consumed;
(3) it stores results in a database
to allow automated analysis of
the vast quantities of debugging data
that large jobs can produce.
GDBase has been used to date
to debug runs of more than 8,000 MPI tasks.
Biography
Dr. Dan Stanzione,
Director of the
High Performance Computing Initiative (HPCI)
at Arizona State University,
joined the
Ira A. Fulton School of Engineering
in 2004.
Prior to ASU,
he served as an AAAS Science Policy Fellow
in the Division of Graduate Education
at the National Science Foundation.
Stanzione began his career at
Clemson University,
where he earned his doctoral and master degrees
in computer engineering
as well as his bachelor of science in
electrical engineering.
He then directed
the supercomputing laboratory at Clemson
and also served as
an assistant research professor
of electrical and computer engineering.
Dr. Stanzione's research focuses on
parallel programming,
scientific computing,
Beowulf clusters,
scheduling in computational grids,
alternative architectures for
computational grids,
reconfigurable/adaptive computing,
and
algorithms for high performance bioinformatics.
Also an advocate of engineering education,
he facilitates student research
through the HPCI
and teaches
specialized computation engineering courses.
Research Scientist
ConocoPhillips
School of Geology & Geophysics
University
of Oklahoma
Topic:
"GEON2
and the OpenEarth Framework (EOF)"
Slides:
PowerPoint
PDF
Talk Abstract: coming soon
Biography: coming soon
Computer Scientist
Software Group (76 SMXG)
Tinker
Air Force Base
Topic:
"Towards a Computationally Bound
Numerical Weather Prediction Model"
Slides:
PowerPoint
PDF
Talk Abstract
Over the past 15 years,
the vector processor based supercomputer
has become virtually extinct in mainstream
High Performance Computing
(HPC).
A few hardware vendors
continue to
supply these types of systems to niche users,
but the larger HPC community,
or at least the systems that
most users have access to,
are married to
price/performance-minded scalar technology.
During this HPC revolution,
the weather community has watched
the efficiency of the weather codes decrease,
in terms of single processor performance,
from nearly 90%
(computationally bound)
to 5%
(memory bound)
of peak floating point performance.
The primary reason is that
the time required to access
Commodity Off-The-Shelf
(COTS)
main memory components
has been outpaced by gains in
CPU clock speed,
core counts
and
vector lengths,
and
therefore in floating point processing speed.
Future multi-core and many-core
technology developments
will exacerbate
the memory boundedness of
existing weather forecast models
on large multi-core HPC solutions.
As a result and
in order to fully utilize
upgrades in computing power,
the meteorological community
has had to consider
the costly exercise of
developing new methods for
computing solutions efficiently on
emerging hardware.
If a weather forecast modeler
could recoup some of the lost efficiency,
one could improve the weather forecast
by providing a higher resolution forecast
in the same amount of wall clock time.
Our work focuses on
what programmers need to consider
in order to achieve
the best possible performance
for their application
—
in other words,
to achieve CPU-bound status.
We will review current solution techniques
used to solve the meteorological equation set,
stress the importance of
reducing the memory footprint of data
and increasing
the ratio of computations to memory accesses.
We will show results from a new model
that uses these methods
with a goal of sustaining
a threefold increase in efficiency
using large multi-core processors.
Biography
Dr.
Dan Weber
has 22 years of experience in
numerical weather prediction and forecasting.
His passion has been
to build and optimize
numerical weather prediction models
for use in weather forecasting.
In addition to
performing research
and
writing numerous papers on thunderstorms
and computer software optimization techniques
targeted at massively parallel computers,
he has taught courses in
weather forecasting techniques
and
severe and unusual weather,
and has held positions with the
National
Weather Service,
at the
University
of Oklahoma
(OU)
and in private industry.
Dr. Weber is currently employed at
Tinker
Air Force Base
and is working to incorporate
complex weather data into
simulation systems
using supercomputers.
Dr. Weber graduated with
undergraduate and graduate degrees in
Meteorology and Geology
from the
University
of Utah
and a doctoral degree in
Meteorology
from OU.
His current research interests include
high-resolution modeling of
thunderstorms,
aircraft turbulence,
and
urban weather,
including airflow around urban structures.
Dr. Weber has participated in
several forensic weather projects
and
has supported the
Korean
Meteorological Administration's
real-time weather forecasting efforts
via the installation and optimization of
a state of the art weather prediction system
that he helped develop at OU.
Assistant Professor
Department
of Computer Science
University of
Arkansas at Little Rock
Topic:
"Mining for Science and Engineering"
Slides:
PowerPoint
PDF
Talk Abstract
Current models of social networks
fail to capture
many of the intricacies of
the rich complex structures
that real world networks exhibit.
These include issues such as
community hierarchies and overlaps,
horizontals across
an organization with
heterogeneous business units,
node and edge attributes,
types,
multiplicity,
location and capacity,
etc.
Most previous work on
the analysis of networks
has focused on
a set of important but relatively simple
measures of network structure.
Through this work,
we propose to focus on
the increasing role of
'mining' / 'using'
the voluminous data
being generated across
several applications
for the pursuit of
an holistic framework of methods,
tools/toolkits and methodologies
(from general-purpose
to domain / application-specific
customizations)
to deal with
the ever increasing amount of data
that drives these applicative needs.
Biography
Kenji
Yoshigoe
is an assistant professor at the
Department
of Computer Science
at the
University of
Arkansas at Little Rock
(UALR).
He received his Ph.D. in
Computer Science and Engineering
from the
University of South Florida.
He serves as a lab manager at
UALR's
High Performance Computing facility,
currently being developed.
His research interest is in
performance evaluation of computer networks
ranging
from high-speed routers
to wireless sensor networks.