1 / 16

Department of Energy High Performance Computing

Department of Energy High Performance Computing. Optical Interconnect Workshop November 8, 1999 George Seweryniak ESnet Program Manager DOE HQ (GTN). . 10 TF. 1 TF. 100 GF. 10 GF. System Architecture Trends. 1999. 2000. Architectures increasing: Processor performance

keely
Download Presentation

Department of Energy High Performance Computing

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Department of Energy High Performance Computing Optical Interconnect Workshop November 8, 1999 George Seweryniak ESnet Program Manager DOE HQ (GTN)

  2. 10 TF 1 TF 100 GF 10 GF System Architecture Trends 1999 2000 • Architectures increasing: • Processor performance • Processors / node, nodes / System • Faster interconnects 2001 # Processors / Nodes # Nodes / System

  3. DOE Challenges Materials Many DOE Programs Need Dramatic Advances in Computing Capabilities To Meet Their Mission Goals Combustion Global Systems Health Effects, Bioremediation Subsurface Transport Fusion Energy

  4. DOE Computing Challenges 1000 Building on a long tradition of driving the supercomputer industry, DOE, via ASCI, is once again fostering dramatic advancements in scientific computing. SSI will make these capabilities available to the broader scientific and engineering communities —at much reduced cost. ASCI Curve 100 Teraflops 10 Moore’s Law “business as usual” 1 0.1 1996 2006 1998 2000 2002 2004

  5. DOE ASCI Challenge

  6. Scientific Simulation Initiative In FY2000 the U.S. Department of Energy requested funding to initiate the Scientific Simulation Initiative. The SSI seeks to: “… revolutionize the way that the Department of Energy solves its most demanding, mission-critical problems by fully utilizing the power of advanced computational simulation.” Ernest J. Moniz Undersecretary of Energy December 1997

  7. Initial SSI Program Combustion Systems Global Systems Basic Sciences • IC Engines • Gas Turbines • Boilers • Burners • Atmosphere • Ocean • Land • Sea Ice • Transport • Genomics • Materials • Fusion • QCD • Nuclear Theory • ... Terascale Systems Software Terascale Computing Infrastructure

  8. Terascale Computing Infrastructure Ultrahigh-speed Networks • 10-... Gbits/sec backbone network • 1-... Gbits/sec network to support major sites • Team with IT2 agencies to extend networking capabilities to other sites Large-scale Computing Systems • A site to operate a leading-edge tera-scale (5-40-… tf) computer system • A terascale computing infrastructure to enable full utilization of the lead-ing-edge capability • Team with other agencies to extend computing capabilities to other sites.

  9. Global Climate Computing

  10. Meteorological Models • Finer grid resolution provides more detail • Increase vertical and horizontal resolution • Data collection can be a major problem • Requires more computational power and storage • Can add more physics to the model • Need high speed interconnects for computation/storage

  11. Combustion Systems Advances in combustion systems simulation requires fundamental advances in a number of areas, including: • Development and utilization of high-resolution models of combustion devices (engines, burners, etc.); • Development of reduced chemical mechanisms for fuel oxidation and pollutant formation; • Development of rigorous science-based descriptions of processes, such as turbulence-chemistry inter-actions, droplet evaporation, etc.; and • Integration of these submodels into a high-fidelity model of the complete system; as well as in using terascale computer systems for the wide range of scientific applications (mechanical engineering, fluid dynamics, chemistry, materials science) involved in combustion. iso-octane

  12. Problem Ensembles

  13. Technical Challenges • Computers • Terascale computers are based on a new architecture —massively parallel arrays of multiprocessors • require very fast / reliable interconnects • Systems Software • Operating system software does not provide all of the needed functionality—lack of vendor financial interest • Applications systems software is non-existent for the scale of problems of interest • Simulation Software • Existing software does notexploit the power of the new architecture • Industry • Government / industry partnerships required to meet the challenges

  14. Can DOE Meet the Challenge? • Because of its mission, DOE has long been a lead agency in high performancecomputing, scientific simulation, computer science, and applied mathematics • DOE has a long and successful history of establishing and operating world-class computing facilities and first-of-a-kind computer systems—from the IBM Stretch in the 1960’s to the massively parallel computers of the 1990’s • DOE has a long and successful history of managing large, complex research projects

  15. DOE Commitment • DOE cannot meet its mission without Terascale computing and Terascale infrastructure • cannot solve it alone • need to work with industry • system level approach

  16. Optical Interconnects • A light at the end of the tunnel??? THE END

More Related