DOC PREVIEW
Berkeley COMPSCI C267 - 21st Century Engines of Discovery

This preview shows page 1-2-3-22-23-24-44-45-46 out of 46 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 46 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

21st Century Engines of DiscoveryLaplace Anticipates Modern High-End ComputersWho Needs High-End Computers?Early Days of Parallel Computing (1985-1995)Where Are We Today?LBNL’s Seaborg SystemApplications for Future Petaflops-Scale ComputersNanoscienceIron-Manganese Runs on SeaborgNanoscience Code CharacteristicsNanoscience RequirementsClimate ModelingClimate Modeling Code CharacteristicsClimate Modeling RequirementsFusion Energy ResearchFusion Code CharacteristicsFusion RequirementsCombustionCombustion Code CharacteristicsCombustion RequirementsAstrophysics and CosmologyAstrophysics Code CharacteristicsAstrophysics RequirementsComputational BiologyBiology ApplicationsBiology Characteristics and RequirementsExperimental Mathematics: Discovering New Theorems by ComputerSome Supercomputer-Class Experimental Math ComputationsResearch Questions for Future High-End Computing40 Years of Moore’s LawBeyond Silicon: Sooner Than You ThinkSelf-Assembled Wires 2nm Wide [P. Kuekes, S. Williams, HP Labs]Rotaxane Molecular Switch [Prof. Fraser Stoddart, UCLA]Fundamental Device LimitsAmdahl’s Law and Little’s LawThe Performance Evaluation Research Center (PERC)Performance ModelingPerformance Modeling: Prediction Versus Measurement“What-If” Analysis for IBM Power 3 System with Colony SwitchPerformance Tools: The PAPI PerformeterImprovement to Climate Modeling Code Using Performance ToolsGeneric Code OptimizationConclusions21st Century Engines of DiscoveryDavid H. BaileyChief Technologist, Computational Research DeptLawrence Berkeley National Laboratoryhttp://crd.lbl.gov/~dhbaileyLaplaceAnticipates Modern High-End Computers“An intelligence knowing all the forces acting in nature at a given instant, as well as the momentary positions of all things in the universe, would be able to comprehend in one single formula the motions of the largest bodies as well as of the lightest atoms in the world, provided that its intellect were sufficiently powerful to subject all data to analysis; to it nothing would be uncertain, the future as well as the past would be present to its eyes.”-- Pierre Simon Laplace, 1773Who Needs High-End Computers?Expert predictions:♦ (c. 1945) Thomas J. Watson (CEO of IBM): “World market for maybe five computers.”♦ (c. 1975) Seymour Cray: “Only about 100 potential customers for Cray-1.”♦ (c. 1977) Ken Olson (CEO of DEC):“No reason for anyone to have a computer at home.”♦ (c. 1980) IBM study:“Only about 50 Cray-1 class computers will be sold per year.”Present reality:♦ Many homes now have 5 Cray-1 class computers.♦ Latest PCs outperform 1990-era supercomputers.Early Days of Parallel Computing (1985-1995)♦ Practical parallel systems became commercially available for the first time.♦ Some scientists obtained remarkable results.♦ Many were enthused at the potential for this technology for scientific computing.BUT♦ Numerous faults and shortcomings of these systems (hardware and software) were largely ignored.♦ Many questionable performance claims were made, both by vendors and by scientific users – “scientific malpractice.”Where Are We Today?♦ Several commercial vendors are producing robust, high-performance, well-supported systems.♦ New systems achieve up to 50 Tflop/s (5 x 1013flops/sec) peak performance, and 5-20 Tflop/s sustained performance on real scientific computations.♦ We are on track to achieve 1 Pflop/s by 2009 or 2010.♦ Many scientists and engineers have converted their codes to run on these systems.♦ Numerous large industrial firms are using highly parallel systems for real-world engineering work.But numerous challenges remain.LBNL’s Seaborg System♦ 6000-CPU IBM SP: 10 Tflop/s (10 trillion flops/sec).♦ Currently #21 on Top500 list of most powerful systems.Applications for Future Petaflops-Scale Computers♦ Protein folding.♦ Weather forecasting.♦ Business data mining.♦ DNA sequence analysis.♦ Experimental mathematics.♦ Inter-species DNA analyses.♦ Medical imaging and analysis.♦ Nuclear weapons stewardship. ♦ Multiuser immersive virtual reality.♦ National-scale economic modeling.♦ Climate and environmental modeling.♦ Molecular nanotechnology design tools.♦ Cryptography and digital signal processing.NanoscienceSimulations of physical phenomena at the nanometer scale lead to future nanotech-produced materials and devices.Iron-Manganese Runs on SeaborgNanoscience Code Characteristics♦ Many-body methods:♦ Quantum Monte-Carlo.♦ Eigenfunction-type calculations (diagonalization of large matrices).♦ Single-particle methods:♦ Wave functions are expanded in plane waves.♦ Large 3-D FFTs, dense linear algebra.♦ Classical molecular dynamics codes:♦ Model inter-molecular forces using classical methods.♦ Used to study synthesis of nanostructures and large structures beyond the scope of quantum calculations.Nanoscience RequirementsElectronic structures and magnetic materials:♦ Current: ~500 atom; 1.0 Tflop/s, 100 Gbyte memory.♦ Future (hard drive simluation): 5000 atom; 30 Tflop/s, 4 Tbyte memory.Molecular dynamics:♦ Current: 109atoms, ns time scale; 1 Tflop/s, 50 Gbyte mem.♦ Future: alloys, us time scale; 20 Tflop/s, 5 Tbyte memory.Continuum solutions:♦ Current: single-scale simulation; 30 million finite elements.♦ Future: multiscale simulations; 10 x current requirements.Climate ModelingLarge community-developed codes, involving complex coupling of atmosphere, ocean, sea ice, land systems, are used to study long-term climate trends.Climate Modeling Code Characteristics♦ Solve equations of hydrodynamics, radiation transfer, thermodynamics, chemical reaction rates.♦ Large finite difference methods, on regular spatial grids (require high main memory bandwidth).♦ Short- to medium-length FFTs are used, although these may be replaced in future.♦ Sea ice and land codes are difficult to vectorize.♦ Scalability is often poor, due to limited concurrency.Scientists would love to use finer grids, which would exhibit greater scalability, but then a century-long simulation would not be feasible.Climate Modeling RequirementsCurrent state-of-the-art:♦ Atmosphere: 1 horizontal deg spacing, with 29 vertical layers.♦ Ocean: 0.25 x 0.25 degree spacing, 60 vertical layers.♦ Currently requires one minute run time per simulated day, on 256 CPUs.Future requirements (to resolve ocean mesoscaleeddies):♦ Atmosphere: 0.5 x 0.5 deg spacing.♦ Ocean: 0.125 x


View Full Document

Berkeley COMPSCI C267 - 21st Century Engines of Discovery

Documents in this Course
Lecture 4

Lecture 4

52 pages

Split-C

Split-C

5 pages

Lecture 5

Lecture 5

40 pages

Load more
Download 21st Century Engines of Discovery
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view 21st Century Engines of Discovery and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view 21st Century Engines of Discovery 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?