| |
|
|
Head-on view of a protein nanotube from prototype NASA Growler
software for visualizing molecular structures of 1,000 or more
atoms.
|
Representative FY2003 agency activities
NSF: Terascale infrastructure; systems software,
middleware, software environments, libraries, visualization, data
management, and algorithms for heterogeneous distributed high-end
systems; grid resource management; quantum and biological concepts
DARPA: Polymorphous architectures; high-end productive,
robust, intelligent computing systems; very largescale integration
of photonics for intra- and inter-chip communication, including
processor-in-memory arrays
DARPA/NSF: Biomolecular structures applied to terascale
computation and storage
NASA: High-end software and systems tuning and
management techniques; Information Power Grid technologies and tools;
information physics (properties of sensing, processing, and storage
systems); quantum and nanoscale technologies
NIH: High-end biomedical computing; tools for determining 3-D
molecular structures; methods for displaying and analyzing images
from instrumentation data
DOE Office of Science: Scalable mathematical
algorithms and software infrastructure (operating systems, component
technologies, optimal mathematical solvers) for terascale modeling
and simulation applications; partnerships for terascale science
DOE/NNSA: Science and engineering innovations in
high-speed computation, terabyte data storage and retrieval, and
visualization to enable supercomputer modeling and simulation for
U.S. nuclear
stockpile stewardship
NSA: Collaborations with high-end systems manufacturers;
operating system and programming language
improvements; fundamental technologies for special-purpose devices
(power controls, cooling, interconnects, switches ,and design tools);
computer memory performance; fundamental physics of quantum information
systems
NOAA: Improved climate and weather models via enhanced
Modular Ocean Model, Flexible Modeling System, and Scalable Modeling
System
NIST: Research in quantum computing, secure quantum
communication, optimization and computational geometry, photonics,
nanotechnologies, optoelectronics, and new chip designs and fabrication
methods
ODDR&E: University-based research in quantum
communications and memory EPA : Paradigm s , techniques, and tools
for modeling complex environmental phenomena such as interactions
of air, water, and soil
|
We are in the early stage of a revolution
in science nearly as profound as the one that occurred early in
the last century with the birth of quantum mechanics.This revolution
is caused by two developments: one is the set of instruments such
as electron microscopy, synchrotron x-ray sources, lasers, scanning
microscopy, and nuclear magnetic resonance devices; the other is
the availability of powerful computing and information technology.
Together these have brought science finally within reach of a new
frontier, the frontier of complexity.
- John H. Marburger III, Director,
OSTP
Beyond this frontier lie the tiniest units
and processes of organic and inorganic matter, the largest structures
and farthest reaches of the universe, the interactions and patterns
of change among the elements of the biosphere that enables life
on Earth, and the possibility of novel materials, pharmaceuticals,
technologies, and tools we can barely imagine today.
The NITRD agencies are aggressively pursuing
technical breakthroughs in component technologies, high-end system
and storage architectures, systems software, and programming environments
that will enable U.S. science to lead the world in these promising
realms. Attaining the necessary highest-performance capabilities
requires fundamental long-term research in computer architecture,
semiconductor design, and systems software, as well as in areas
with revolutionary potential, such as quantum and biomolecular concepts,
in all aspects of high-end systems and software (see Terascale
Infrastructure for Discovery). NITRD research provides an essential
bridge between the requirements of the Federal government for cutting-edge
national defense, national security,and scientific applications
and commercially available computing products.
In addition, NITRD-funded research is developing technologies and
tools for high-end collaboratories - "virtual" lab facilities
shared over high-performance networks by distributed teams of scientists.
The goal is a networked infrastructure for advanced research that
can seamlessly connect distributed teams to the high-end computing
systems, instruments, advanced simulation and visualization software,
and sensor networks they need to work collaboratively on data- and
computation-intensive problems.
One form of such interconnections is called
"grid computing" because it works as a patterned overlay
on advanced networks, creating a grid of linked, interoperating
resources that can be widely distributed, with a governing fabric
of protocols and protections for the grid and its users as a whole.
In effect, a grid leverages the Internet to provide some of the
infrastructure for large-scale, distributed, high-performance computing,
and it extends high-end capabilities to a wider community of scientists,
engineers , and students.
With early, visionary support from DARPA,
DOE, NASA, and NSF over the past five years, researchers at the
leading edge of grid design prototyped and experimented with the
open-standard Globus Toolkit. Developed at DOE's Argonne National
Laboratory (ANL) and the University of Southern California, Globus
is the first suite of middleware software tools for linking and
using computing resources in a grid environment. NASA adopted Globus
to build its Information Power Grid, an ambitious initiative to
network the agency's high-end computing resources and data repositories.
In FY 2002, NSF brought national visibility to the grid concept
when it awarded an unprecedented $53 million to four institutions
to create a "Distributed Terascale Facility" with Globus
as the underpinning that knits the components into a whole (story
on next page).
In recent months, 12 leading hardware
and software vendors - including Compaq, Cray, Fujitsu, Hitachi,
IBM, Microsoft, NEC, SGI, and Sun Microsystems - have announced
that they will adopt the Globus Toolkit as the basis for grid
computing with their products. An IBM executive has predicted that
grid computing will transform U.S. business practices.
Even as NITRD research generates many
such promising technology transfers, the Program's primary focus
in high-end computing continues to be on research to increase the
most advanced and exacting capabilities of high-end platforms for
critical U.S. defense and civilian requirements, including global
science and technology leadership. For FY 2003, the NITRD agencies
will support research in: 1) the fundamental science of supercomputing
component technologies and system designs, and 2) technologies,
tools, and applications to enable the Nation's researchers to work
computationally at the state of the art.
Major Research Challenges
- High performance computer architecture - understanding and managing
the complexity tradeoffs among hardware design and architecture,
systems software, and scientific applications to deliver the greatest
capability for scientific discovery and national security
- Revolutionary approaches - such as innovative processing and
storage concepts, including novel architectures, quantum and biomolecular
components, hybrid technologies, reconfigurable systems on a chip,
and processor-in-memory (PIM) technologies - to enable nextgeneration
supercomputing platforms
- Systems software - advanced programming environments, compilers,
libraries, middleware, and performance engineering technologies
- New high-end algorithms and codes for scientific computation
and simulation; integrated and optimized software infrastructure
for distributed terascale computing environments
|