Docstoc

BSC

Document Sample
BSC Powered By Docstoc
					                 Barcelona
                 Supercomputing
                 Center
                 Centro Nacional de Supercomputación



Prof. Mateo Valero
BSC Director



                                       Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

               What drives HPC ? --- “The Need for Speed...”
     Computational Needs of Technical, Scientific, Digital Media and Business Applications
                        Approach or Exceed the Petaflops/s Range
                                     CFD Wing Simulation
                            512x64x256 Grid
                                                                                                          Materials Science
                            (8.3 x10e6 mesh points)
                            5000 FLOPS per mesh point,
                                                                                              Magnetic Material:
                            5000 time steps/cycles
                                                                                               Current: 2000 atoms; 2.64 TF/s, 512 GB
                            2.15x10e14 FLOPS
                                                                                               Future: HDD Simulation - 30 TF/s, 2 TB
                                   CFD Full Plane Simulation                                  Electronic Structures:
                            512x64x256 Grid                                                    Current: 300 atoms; 0.5 TF/s, 100 GB
                            (3.5 x10e17 mesh points)                                           Future: 3000 atoms; 50 TF/s, 2TB
                            5000 FLOPS per mesh point,
                            5000 time steps/cycles
                            8.7x10e24 FLOPS                                                                          Source: D. Balley, NERSC
Source: A. Jameson, et al
                                                                                                  Spare Parts Inventory Planning

                              Digital Movies and Special Effects                              Modelling the optimized deployment of
                                                                                              10000 part numbers across 100 part
                            ~1e14 FLOPs per frame                                             depots and requries:
                            50 frames/sec                                                     - 2x10e14 FLOP/s
                            90 minute movie                                                     (12 hours on 10, 650 MHz CPUs)
                             - 2.7e19 FLOPs                                                   - 2.4 PetaFlop/s sust. performance
                                                                                                (1 hour turn-around time)
                            ~ 150 days on 2000 1GFLOP/s CPUs                                  Industry trend to rapid, frequent modeling
                                                                                              for timely business decision support driver
       Source: Pixar                                               Source: B. Dietrich, IBM
                                                                                              higher sustained performance
                                                                                                         Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Technological Achievements

     •   Transistor (Bell Labs, 1947)

         •   DEC PDP-1 (1957)

         •   IBM 7090 (1960)

     •   Integrated circuit (1958)

         •   Kilburn, TI/Fairchaild

         •   IBM System 360 (1965)

         •   DEC PDP-8 (1965)

     •   Microprocessors (1971)

         •   Intel 4004



                                        Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   In 50++ years ...




                            Power5, 2004
                            First dual core SMT
                            processor
                            276M transistores
        Eniac, 1946 ...
                            389 mm2
        1900 vacuum tubes
                            1,8 GHz           Barcelona, 16 de Noviembre de 2004
                            1,9 MB L2 cache
BSC - Barcelona Supercomputer Center

   Technology Directions: SIA Roadmap

           Year          1999    2002   2005   2008   2011        2014
     Feature size (nm)   180     130    100     70     50            35

     Logic trans/cm2     6.2M    18M    39M    84M    180M         390M

     Cost/trans (mc)     1.735   .580   .255   .110   .049         .022

     #pads/chip          1867    2553   3492   4776   6532         8935

     Clock (MHz)         1250    2100   3500   6000   10000       16900

     Chip size (mm2)     340     430    530    620     750          900

     Wiring levels        6-7     7      7-8    8-9     9            10

     Power supply (V)     1.8     1.5    1.2    0.9    0.6          0.5
     High-perf pow
                          90     130    160    170     175          183
     (W)
     Battery pow (W)      1.4     2      2.4    2.8    3.2          3.7




                                                              Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Applications for Supercomputers

     •   Aircraft/car simulations

     •   Life Science (Proteins, Human Organs,…)

     •   Atmosphere

     •   Stars

     •   Nanomaterials

     •   Drugs

     •   Regional/Global Climate/Weather/Pollution

     •   High Energy Physics

     •   Combustion

     •   Image Processing
                                                     Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center
    Suitable applications for massively parallel
    systems




     Source: Rick Stevens,
   Argonne National Lab and
   The University of Chicago                   Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Throughput vs. Parallel programming

     •   Throughput

         •   Multiple, unrelated, instruction streams (programs) that execute concurrently on
             multiple processors

         •   Multiprogramming n tasks on p processors: each task receives p/n processors

     •   Parallel Programming

         •   Multiple related, interacting instructions (single program) that execute
             concurrently to increase the speed of a single program

         •   1 task on p processors, each processor receives 1/p of the task: reduce
             response time




                                                                                 Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

      Distributed and Parallel Systems




 Distributed                                                   Massively parallel
 systems                                                       systems
 heterogenous                                                  homogenous



     ● Gather (unused) resources            ● Bounded set of resources
     ● Steal cycles                         ● Apps grow to consume all cycles
     ● System SW manages resources          ● Application manages resources
     ● System SW adds value                 ● System SW gets in the way
     ● 10% - 20% overhead is OK             ● 5% overhead is maximum
     ● Resources drive applications         ● Apps drive purchase of equipment
     ● Time to completion is not critical   ● Real-time constraints
     ● Time-shared                          ● Space-shared
     ● SETI@home                            ● Blue Gene/L
       ≈ 400,000 machines                     32768 processors
       Averaging 40 Tflop/s                   Averaging 70,72 Tflop/s

                                                             Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center



       •    Google query attributes
            •    150M queries/day (2000/second)
            •    3B document in the index
       •    Data centers
            •    15,000 Linux systems in 6 data centers
                 •    15 TFlop/s and 1000TB total capability
                 •    40-80 1U/2U servers/cabinet
                 •    100 MB Ethernet switches with Gigabit Ethernet uplink
            •    growth from 4,000 systems (June 2000)
                 •    18M queries then
       •    Performance and operation
            •    simple reissue of failed commands to new servers
            •    no performance debugging
                 •    problems are not reproducible
   Source: Monika Henzinger, Google


                                                                              Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Life Science future projects

     •   In silico drug screening

     •   Clustering of Expressed Sequence Tags

     •   In silico Structural Genomics

     •   First ab-initio Calculations of the Electronic Structure of the Smallest Living
         Organism

     •   Mixed QM/MM Simulations of the Five Most Important Enzymatic Reactions

     •   Virtual Cell Project

     •   Digital human and virtual chirurgy

     •   Ab-initio Food Science

     •   Protein Folding Dynamics

                                                                           Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Better understanding of biological process

     •   The structure and dynamics of
         protein folding determine their
         biological properties

     •   Some illness related to misfoldings
         are

         •   Alzheimer’s Disease

         •   Mad Cow Disease

         •   Cystic Fibrosis

     •   Supercomputing will enable the in-
         silico simulation of protein folding,
         contributing to novel drug
         development techniques and new
         medical treatments

                                                 Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Protein Folding Calculation

              Description                 Count                 Comment
      Atoms                                ≈ 32000     260 amino acid protein + water

      Force evaluation / time step           109       Pairwise atom-atom interaction

      FLOPs / force evaluation               150       Typical molecular dynamics

      FLOPSs / time step                  1.5 x 1011

      Each time step                      ≈ 10-15 s    0.5 - 5 femto second

      Total simulation time single step    10-3 s      Protein folds in ≈1 mili second

      Total time steps                    2 x 1011

      FLOPs / simulation                  3 x 1022     Total FLOPs to fold a protein

      Execution time                      3 x 107 s    1 year

      Required FLOPS                       ≈ 1015      1 Petaflop




                                                                                 Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center
   Personalized medicine requires massive
   supercomputing resources
     •   Each individual has a different genetic sequence. Its understanding would
         help to identify specific risks

     •   Sequencing the DNA of the human population implies the analysis of 1.8
         million TB/s (300 TB/person * 6 billion person)

     •   The comparison of the genetic profiles of an individual against a central
         database would allow the development of life-long rational healthcare with
         food and drugs adequate to the individual genetic characteristics




                                                                        Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Protecting and improving our environment

     •   Supercomputing facilitates accelerated and
         significant progress in the earth and material
         sciences

         •   Biogeochemical cycles
         •   Biodiversity and ecosystems in equilibrium
         •   Climatic variations
         •   Hydrological predictions
         •   Contagious diseases and environmental impact
         •   Rational use of resources
         •   Reinventing materials and their applications

     •   Supercomputing will enable to simulate the
         environmental and social impact of agricultural,
         industrial and urbanistic policies, as well as the
         impact of external factors (desertification,
         catastrophic events) in real time

                                                              Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

    CEPBA: Service

                                            Memory         Peak
   Manuf.        Model        Processor                                       Service
                                             Disc      Performance

                              128 Power3    64+18 GB                            10/01
     IBM       SP3+p360                                336 Gflop/s
                              + 36 Power4    1,8 TB                             12/02
              Alpha Server         16         8 GB
   Compaq                                              23,3 Gflop/s             9/00
                GS-160        Alpha 21264    108 GB
                                   16         1 GB
   Parsytec     CCi-8D                                 3,2 Gflop/s              4/98
                               Pentium II     30 GB
    Silicon                        64         8 GB
              Origin 2000                               25 Gflop/s              1/97
   Graphics                  MIPS R10000     360 GB
              Alpha Server         12         2 GB
    Digital                                            10,5 Gflop/s             12/96
                  8400        Alpha 21164     32 GB
    Silicon      Power           12+12        4 GB
                                                                            7/96 - 7/97
   Graphics    Challenge     R8000/R10000     50 GB
   Thinking                       2048       256 MB
                 CM-2                                  640 Mflops           4/92 - 2/98
   Machine                        1 bit
                                             1 GB
   Convex       C3480             8                    0,4 Gflop/s         10/91 - 1/98
                                             16 GB
                                 32         128 MB
   Parsys      SN 1000                                  64 Mflops           5/90 - 5/95
                                T800         2 GB
                                                                    Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   CEPBA: European Mobility Programs

     •   Joint CEPBA - CESCA projects

     •   Stays and access to resources




                   Project            Period     Funding    Visitors

                   HCM               1993-1997        950          112
                   PECO              1995-1996        160              17
                    TMR              1996-2000        935          133
                    IHP              2000-2003        700          142
                             Total                   2745          404




                                                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   CEPBA: R&D projects




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   CEPBA: R&D projects




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   CEPBA: Management of Technology Transfer




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   The Consortium

   •   Consortium includes

       •   Spanish Government (MEC)

       •   Catalonian Government (DURSI)

       •   Technical University of Catalunya

   •   To start operations on January, 2005

   •   Location

       •   c/ Jordi Girona 31
       •   08034 Barcelona




                                               Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Motivation



     •   Significant         contribution   to
         advancement of Science in Spain,
         enabling supercomputing capacity,
         scientific-technical synergies, and
         cost        rationalization thanks to
         economies of scale

     •   Powerful tool to assist research and
         development centers,      public and
         private, generating impulses for a
         new technological environment




                                                 Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Mission

   •”Investigate, develop and manage technology to facilitate the advancement of
   science”




                                                                      Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Objectives

     •   Research in Supercomputing and Computer Architecture

     •   Collaborate in R&D e-Science projects with prestigious scientific teams

     •   Manage BSC supercomputers to accelerate relevant contributions to
         research areas where intensive computing is an enabling technology




                                                                        Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Structure




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   IT research and development projects

     •   Continuation of CEPBA (European Center for Parallelism in Barcelona)
         research lines

     •   Deep Computing

         •   Performance Tools

         •   Parallel programing

         •   Grid

         •   Code Optimization

     •   Computer Architecture

         •   Vector processors

         •   Network processors


                                                                     Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   e-Science projects

     •   R&D collaborations

     •   Computational Biology

     •   Computational Chemistry

     •   Computational Physics

     •   Information based Medicine




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Management projects

     •   Supercomputer Management
         •   System Administration

         •   Users support

     •   Business Development
         •   External Relations

         •   Promotion

         •   Technology Transfer

         •   Education

     •   Administration
         •   Accounting and Finances

         •   Human Resources


                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   MareNostrum

     •   PowerPC 970 FX processors (dual processors)

     •   4GB ECC 333 DDR memory per node

     •   3 networks

         •   Myrinet

         •   Gigabit

         •   10/00 Ethernet

     •   Diskless network support

     •   Linux cluster




                                                       Barcelona, 16 de Noviembre de 2004
 BSC - Barcelona Supercomputer Center

          MareNostrum: Overall system description

27 Compute Racks (RC01-RC27)
                                                                  4 Myrinet Racks (RM01-RM04)
• 162 BC chassis w/OPM and gigabit ether switch
                                                                  • 10 clos256+256 myrinet switches
• 2268 JS20+ nodes w/myrinet daughter card
                                                                  • 2 Myrinet spines 1280s




                                                                                  1 Operations Rack (RH01)
7 Storage Server Racks (RS01-RS07)                                                • 7316-TF3 display
• 40 p615 storage servers 6/rack                                                  • 2 p615 mgmt nodes
• 20 FastT 100 3/rack                                                             • 2 HMC model 7315-CR2
• 20 EXP100 3/rack                                                                • 3 Remote Async Nodes
                                                                                  • 3 Cisco 3550
                                                                                  • 1 BC chassis (BCIO)


                                     1 Gigabit Network Racks
                                     • 1 Force10 E600 for Gb network
                                     • 4 Cisco 3550 48-port for 10/100 network
                                                                                        Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Processor: PowerPC 970FX




                                       Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Blade center racks

     •   6 Blade Centers per rack
                                                    Blade Center

     •   27 racks + 1 Blade Center

     •
                                                    Blade Center
         Cabling per rack

         •   84 fiber cables to myrinet switch      Blade Center

         •   6 Gb to Force10 E600
                                                    Blade Center
         •   6 10/100 cat5 to Cisco

                                                    Blade Center


                                                    Blade Center




                                                 Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Myrinet racks

     •   10 Clos 256x256 switches

         •   Interconnect up to 256 Blades

         •   Connect to Spine (64 ports)

     •   2 Spine 1280

         •   Interconnect up to 10 Clos 256x256
             switches

     •   Monitoring using 10/100 connection




                                                  Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Gb Subsystem: Force 10 E600

     •   Interconnection of Blade Centers

     •   Used for system boot of every blade center

     •   212 internal network cables

         •   170 for blades

         •   42 for p615

     •   76 connection available to external connection




                                                          Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   Scientific Applications




                        Source: Bob Voigt , NSF   Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center

   MareNostrum: number comparison

                                 Project                   Earth     MareNostru
                                           BlueGene /L
                                Columbia                 Simulator       m

          Number of racks                     >64          640          40

       Number of processors      10240        128k         5120        4500

      System peak performance    65 TF       360 TF       41 TF        40 TF

        Total system Memory                  16 TB        10 TB        9 TB

              Power                         < 1.6 MW     5.1 MW       600 KW

            Floor space                      400 m2      3250 m2      120 m2

       Bisectional bandwidth                0.7 TB/s     7.8 TB/s     0.6 TB/s

         Number of cables                     5000        83200        2600




                                                                      Barcelona, 16 de Noviembre de 2004
 BSC - Barcelona Supercomputer Center




Thank you !
Prof. Mateo Valero
BSC Director




                                        Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center




                                       Barcelona, 16 de Noviembre de 2004
BSC - Barcelona Supercomputer Center




                 Para ver esta película, deb e                  Para ver esta película, deb e
                disponer de QuickTime™ y de                    disponer de QuickTime™ y de
            un descompresor DV/DVCPRO - NTSC.              un descompresor DV/DVCPRO - NTSC.




     From July 7th to October 20th               From November 27th to December 7th
  1 image every hour on working days                  1 image every 30 seconds




                                                                              Barcelona, 16 de Noviembre de 2004

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:12
posted:9/30/2012
language:Unknown
pages:53