Early Systems Evaluation at ORNL (green text on poster)
Impact of ORNL Mathematics and Algorithm Development Activities (purple text on poster)
The Problem:
The scientific problems of interest to DOE require creation of realistic
models of physical situations. Realistic models of such challenging problems
require computational power in the 5-10 teraflops
range (and beyond).
With a rich history of evaluating new architectures and
bringing them to production status, ORNL is ready for the challenges of
teraflops computing. The extensive ORNL expertise in tool development
assures readiness for teraflops, whether achieved via "big iron" or
with distributed systems.
ORNL Solutions: ORNL is pursuing the development of the multi-teraflops computing environment needed to address major scientific challenges. ORNL's strategy includes both the detailed evaluation of new architectures, and the development of tools facilitating distributed computing.
Check it out:
|
ORNL HPCC history (timeline details) Early Systems Evaluation at ORNL (green text on poster) Impact of ORNL Mathematics and Algorithm Development Activities (purple text on poster) |
 SRC evaluation plans |
ORNL's Center for Computational Sciences plans to purchase two SRC 6 units for an evaluation process, which will cover systems and components as well as methods to connect multi-processor units. This evaluation process will be a joint effort between ORNL and SRC. "This project shows the Department of Energy's commitment to exploring the frontiers of computing," said Martha Krebs, director of DOE's Office of Energy Research. "A crucial part of this evaluation is determining the effectiveness of this type of supercomputer for solving scientific problems critical to DOE missions. Oak Ridge National Laboratory's experience under DOE's high-performance computing program positions it to effectively carry out this work." |
 Tera MTA |
C-Plant |
As part of our project to evaluate new computer architectures, the evaluation
team is studying the performance of a variety of benchmarks on new parallel
computers. Preliminary work has been done on the Tera MTA machine at SDSC,
Sandia's Computational Plant (
C-plant
), the Swiss T0 (T-zero) machine
at EPFL, and the
Dell Precision 610 dual-processor Xeon system.
Swiss-T0
|
 Single electron devices consist of metal or semiconductor islands (quantum dots), a few nanometers in diameter, surrounded by an insulating moat. These devices can be manipulated to operate as transistors or memory elements. |
Researchers at the Center for Engineering Science and Advanced Research (CESAR) in ORNL's Computer Science and Mathematics Division are currently exploring quantum dot arrays as a promising hardware technology for massively parallel computation. Use of this technology is expected to be invaluable, because of the very low power consumption and ultrahigh density in a chip implementation. Ongoing efforts include:
|
 source: APS |
Quantum communication takes advantage of a fundamental feature
of entangled photons--the instantaneous transfer (teleportation) of the
quantum state of one particle to another independent of distance
between the particles. The Center for
Engineering Science and Advanced Research (CESAR) in ORNL's Computer Science
and Mathematics Division
has established a new laboratory for optical computing and quantum
communications.
|
