Review & Approval System - Search Detail https://cfwebprod.sandia.gov/cfdocs/RAA/templates/index.cfm Search Detail Submittal Details Docum ent I nfo Title : Reversible Logic for Supercom puting Docum ent Num ber : 5 2 3 1 9 0 9 SAND Num ber : 2 0 0 5 -2 6 8 9 C Review Type : Electronic Status : Approved Sandia Contact : DEBENEDI CTI S,ERI K P. Subm ittal Type : Conference Paper New Search Requestor : DEBENEDI CTI S,ERI K P. Subm it Date : 0 4 / 2 2 / 2 0 0 5 Refine Search Search Results Peer Review ed? : N Author( s) Clone Request DEBENEDI CTI S,ERI K P. Edit Request Cancel Request Event ( Conference/ Journal/ Book) I nfo Nam e : Com puting Frontiers 2 0 0 5 City : I schia State : Country : I taly Start Date : 0 5 / 0 4 / 2 0 0 5 End Date : 0 5 / 0 6 / 2 0 0 5 Partnership I nfo Partnership I nvolved : No Partner Approval : Agreem ent Num ber : Patent I nfo Scientific or Technical in Content : Yes Technical Advance : No TA Form Filed : No SD Num ber : Classification and Sensitivity I nfo Title : Unclassified-Unlim ited Abstract : Docum ent : Unclassified-Unlim ited Additional Lim ited Release I nfo : None. DUSA : None. Routing Details Role Routed To Approved By Approval Date Derivative Classifier Approver 0 4 / 2 2 / 2 0 0 5 SUMMERS,RANDALL M. SUMMERS,RANDALL M. Conditions: 1 of 2 1/2/2008 9:23 PM
Review & Approval System - Search Detail https://cfwebprod.sandia.gov/cfdocs/RAA/templates/index.cfm Classification Approver 0 4 / 2 5 / 2 0 0 5 W I LLI AMS,RONALD L. W I LLI AMS,RONALD L. Conditions: Manager Approver 0 4 / 2 9 / 2 0 0 5 PUNDI T,NEI L D. PUNDI T,NEI L D. Conditions: Adm inistrator Approver LUCERO,ARLENE M. KRAMER,SAMUEL 0 6 / 0 5 / 2 0 0 7 Created by W ebCo Problems? Contact CCHD: by em ail or at 8 4 5 -CCHD (2243). For Review and Approval process questions please contact the Application Process Ow ner 2 of 2 1/2/2008 9:23 PM
SAND 2005-2689C Reversible Logic for Supercomputing How to save the Earth with Reversible Computing Erik P. DeBenedictis Erik P. DeBenedictis Sandia National Laboratories May 5, 2005 Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.
Applications and $100M Supercomputers System Applications Applications Technology Performance No schedule provided by No schedule provided by source source Plasma Fusion � Nanotech + Simulation Reversible Logic μ P 1 Zettaflops [Jardin 03] Full Global Climate (green) best-case logic 100 Exaflops [Malone 03] (red) � MEMS 10 Exaflops Optimize 1 Exaflops Compute as fast ↑ � Architecture: IBM 100 Petaflops as the engineer Cyclops, FPGA, PIM can think 10 Petaflops [NASA 99] 1 Petaflops ↑ � Red Storm/Cluster ↓ 100 × ↑ 1000 × [SCaLeS 03] 100 Teraflops Year � 2000 2000 2010 2010 2020 2020 2000 2010 2020 2030 [Jardin 03] S.C. Jardin, “Plasma Science Contribution to the SCaLeS Report,” Princeton Plasma Physics Laboratory, PPPL-3879 UC-70, available on Internet. [Jardin 03] S.C. Jardin, “Plasma Science Contribution to the SCaLeS Report,” Princeton Plasma Physics Laboratory, PPPL-3879 UC-70, available on Internet. [Malone 03] Robert C. Malone, John B. Drake, Philip W. Jones, Douglas A. Rotman, “High-End Computing in Climate Modeling,” contribution to SCaLeS report. [Malone 03] Robert C. Malone, John B. Drake, Philip W. Jones, Douglas A. Rotman, “High-End Computing in Climate Modeling,” contribution to SCaLeS report. [NASA 99] R. T. Biedron, P. Mehrotra, M. L. Nelson, F. S. Preston, J. J. Rehder, J. L. Rogers, D. H. Rudy, J. Sobieski, and O. O. Storaasli, “Compute as Fast as the Engineers Can Think!” [NASA 99] R. T. Biedron, P. Mehrotra, M. L. Nelson, F. S. Preston, J. J. Rehder, J. L. Rogers, D. H. Rudy, J. Sobieski, and O. O. Storaasli, “Compute as Fast as the Engineers Can Think!” NASA/TM-1999-209715, available on Internet. NASA/TM-1999-209715, available on Internet. [SCaLeS 03] Workshop on the Science Case for Large-scale Simulation, June 24-25, proceedings on Internet a http://www.pnl.gov/scales/. [SCaLeS 03] Workshop on the Science Case for Large-scale Simulation, June 24-25, proceedings on Internet a http://www.pnl.gov/scales/. [DeBenedictis 04], Erik P. DeBenedictis, “Matching Supercomputing to Progress in Science,” July 2004. Presentation at Lawrence Berkeley National Laboratory, also published as [DeBenedictis 04], Erik P. DeBenedictis, “Matching Supercomputing to Progress in Science,” July 2004. Presentation at Lawrence Berkeley National Laboratory, also published as Sandia National Laboratories SAND report SAND2004-3333P. Sandia technical reports are available by going to http://www.sandia.gov and accessing the technical library. Sandia National Laboratories SAND report SAND2004-3333P. Sandia technical reports are available by going to http://www.sandia.gov and accessing the technical library.
Objectives and Challenges • Could reversible computing have a role in solving important problems? – Maybe, because power is a limiting factor for computers and reversible logic cuts power • However, a complete computer system is more than “low power” – Processing, memory, communication in right balance for application – Speed must match user’s impatience – Must use a real device, not just an abstract reversible device
Outline • An Exemplary Zettaflops Problem • The Limits of Current Technology • Arbitrary Architectures for the Current Problem – Searching the Architecture Space – Bending the Rules to Find Something – Exemplary Solution • Conclusions
Simulation of Global Climate Stott et al, Science 2000 “Simulations of the response to natural forcings alone … do not explain the warming in the second half of the century” “..model estimates that take into account both greenhouse gases and sulphate aerosols are consistent with observations over this*period” - IPCC 2001
FLOPS Increases for Global Climate Issue Scaling 1 Zettaflops Ensembles, scenarios Embarrassingly 10 × Parallel 100 Exaflops Run length Longer Running 100 × Time 1 Exaflops New parameterizations More Complex 100 × Physics 10 Petaflops Model Completeness More Complex 100 × Physics 100 Teraflops Spatial Resolution Resolution 10 4 × (10 3 × -10 5 × ) 10 Gigaflops Clusters Now In Use (100 nodes, 5% efficient) Ref. “High-End Computing in Climate Modeling,” Robert C. Malone, LANL, John B. Drake, ORNL, Philip W. Jones, LANL, and Douglas A. Rotman, LLNL (2004)
Outline • An Exemplary Zettaflops Problem • The Limits of Current Technology • Arbitrary Architectures for the Current Problem – Searching the Architecture Space – Bending the Rules to Find Something – Exemplary Solution • Conclusions
Scientific Supercomputer Limits Best-Case Microprocessor Physical Source of Logic Architecture Factor Authority Reliability limit Esteemed physicists 2 × 10 24 logic ops/s 750KW/(80k B T) (T=60°C junction temperature) Derate 20,000 convert Floating point engineering logic ops to floating point (64 bit precision) Expert 100 Exaflops 800 Petaflops Derate for manufacturing Estimate Opinion � 125:1 � margin (4 × ) Estimate 25 Exaflops 200 Petaflops Uncertainty (6 × ) Gap in chart 4 Exaflops 32 Petaflops Improved devices (4 × ) Estimate 1 Exaflops 8 Petaflops Projected ITRS ITRS committee of experts improvement to 22 nm (100 × ) Assumption: Supercomputer 80 Teraflops is size & cost of Red Storm: Lower supply voltage ITRS committee of experts US$100M budget; consumes (2 × ) 2 MW wall power; 750 KW to active components 40 Teraflops Red Storm contract
Outline • An Exemplary Zettaflops Problem • The Limits of Current Technology • Arbitrary Architectures for the Current Problem – Searching the Architecture Space – Bending the Rules to Find Something – Exemplary Solution • Conclusions
Supercomputer Expert System Application/Algorithm run time model as in applications modeling Results Expert System & 1. Block diagram Logic & Memory Technology Optimizer picture of optimal design rules and performance (looks for best 3D system (model) parameters for various mesh of 2. Report of technologies generalized MPI FLOPS count as (CMOS, Quantum Dots, connected nodes, a function of C Nano-tubes …) μ P and other) years into the future Interconnect Speed, power, pin count, etc. Time Trend Physical Lithography as a Cooling, packaging, function of years etc. into the future
Sample Analytical Runtime Model • Simple case: finite • Volume-area rule – Computing ∝ difference equation n 3 • Each node holds n × n × n – Communications ∝ n 2 grid points T step = 6 n 2 C bytes T byte + n 3 F grind /floprate Face-to-face Volume n n 2 cells n 3 cells n n
Expert System for Future Supercomputers • Applications Modeling • Use “Expert System” To – Runtime Calculate: T run = f 1 (n, design) Max n: $<C 1 , T run <C 2 • Technology Roadmap All designs – Gate speed = f 2 (year), – chip density = f 3 (year), • Report: – cost = $(n, design), … • Scaling Objective Function Floating operations – I have $C 1 & can wait T run (n, design) T run =C 2 seconds. What is the biggest n I can and illustrate “design” solve in year Y?
Recommend
More recommend
