DOCSLIB.ORG

Performance Modeling the Earth Simulator and ASCI Q

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Performance Modeling the Earth Simulator and ASCI Q PAL CCS-3 Performance Modeling the Earth Simulator and ASCI Q Darren J. Kerbyson Adolfy Hoisie, Harvey J. Wasserman Performance and Architectures Laboratory (PAL) Los Alamos National Laboratory April 2003 Los Alamos PAL CCS-3 “26.58Tflops on AFES … 64.9% of peak (640nodes)” “14.9Tflops on Impact-3D …. 45% of peak (512nodes)” “10.5Tflops on PFES … 44% of peak (376nodes)” 40Tflops 20Tflops Los Alamos PAL Talk Overview CCS-3 G Overview of the Earth Simulator – A (quick) view of the architecture of the Earth Simulator (and Q) – A look at its performance characteristics G Application Centric Performance Models – Method of comparing performance is to use trusted models of applications that we are interested in, e.g. SAGE and Sweep3D. – Analytical / Parameterized in system & application characteristics G Models can be used to provide: – Predicted performance prior to availability (hardware or software) – Insight into performance – Performance Comparison (which is better?) G System Performance Comparison (Earth Simulator vs ASCI Q) Los Alamos PAL Earth Simulator: Overview CCS-3 ... ... RCU RCU . RCU AP AP AP Memory Memory Memory 0-7 0-7 0-7 ... Node0 Node1 Node639 640x640 ... crossbar G 640 Nodes (Vector Processors) G interconnected by a single stage cross-bar – Copper interconnect (~3,000Km wire) G NEC markets a product – SX-6 – Not the same as an Earth Simulator Node - similar but different memory sub-system Los Alamos PAL Earth Simulator Node CCS-3 Crossbar LAN network Disks AP AP AP AP AP AP AP AP 0 1 2 3 4 5 6 7 RCU IOP Node contains: 8 vector processors (AP) 16GByte memory Remote Control Unit (RCU) I/O Processor (IOP) 2 3 4 5 6 0 1 31 M M M M M M M . M G Each AP has: – A vector unit (8 sets of 6 different vector pipelines), and a super-scalar unit – 5,000 pins – AP operates mostly at 500MHz: – Processor Peak Performance = 500 x 8 (pipes) x 2 (float-point) = 8,000 Mflop/s – Memory bandwidth = 32 GB/s per Processor (256GB/s per node) Los Alamos PAL ASCI Q CCS-3 G 2048 Nodes Node G Fat-tree network (Quadrics) CPU3 CPU PCI PCI G Each node is an HP ES45 2 x2 x2 CPU – 4 x Alpha EV68 micro-processors 1 – 1.25GHz CPU 0 M0 Switch – Memory Hierarchy: L1 64KB I+D Based » 64KB I+D L1 Inter-connect L2 16MB M1 » 16MB L2 cache » 16 GB memory per node (typical) Los Alamos PAL CCS-3 Earth Simulator ASCI Q (similar to NEC SX-6) (HP ES45) Node Architecture Vector SMP Microprocessor SMP System Topology Crossbar (single- Fat-tree stage) Number of nodes 640 2048 Processors - per node 8 4 - system total 5120 8192 Processor Speed 500 MHz 1.25 GHz Peak speed - per processor 8 Gflops 2.5 Gflops - per node 64 Gflops 10 Gflops Memory - per node 16 GB 16 GB (max 32 GB) - per processor 2 GB 4 GB (max 8 GB) Memory Bandwidth (peak) - L1 Cache N/A 20 GB/s - L2 Cache N/A 13 GB/s - Main memory (per PE) 32 GB/s 2 GB/s Inter-node MPI communication - Latency 8.6 sec 5 sec - Bandwidth 11.8 GB/s 300 MB/s Los Alamos PAL A quotation from a CCS-3 scientific explorer? “Its life Jim, but not as we know it” “Its performance Jim, but not as we expected it” Los Alamos PAL Need to have an expectation CCS-3 G Complex machines and software – single processors, interactions within nodes, interaction between nodes (communication networks), I/O G Large cost for development, deployment and maintenance G Need to know in advance what performance will be. Update of SW and/or HW Maintenance Verification of ASCI Q Performance Installation A Model Procurement What should we buy? is the Key (ASCI Purple) Implementation Some measurement possible (small scale) Design Lots of system choices Los Alamos PAL CCS-3 How many _ ? ( How many Alphas are equivalent to the Earth Simulator ? ) Answer: Earth Simulator = 640 x 8 x 8 = 40-Tflops Nodes PEs/node Gflops/PE Alpha ES45 System = ? x 4 x 2.5 = 40-Tflops -> ? = 4,096 nodes Another Answer: It depends on what is going to run Los Alamos PAL Performance Model for SAGE CCS-3 G Identify and understand the key characteristics of code G Main Data structure decomposition: Slab PE – communications scale as number of PEs (^2/3) 1 2 – communication distance (PEs) increases 3 4 – Effect of network topology G Processing Stages – Gather data, Computation, Scatter Data n-4 n-3 n-2 n-1 n n+1 n+2 n+3 n+4 Gather Gather/Scatter Comms Direction Size (cells) X, HI 4 Computen-4 n-3 n-2 n-1 n n+1 n+2 n+3 n+4 X, LO 4 Y, LO 152 Y, HI 152 Z, LO 6860 Scatter n-4 n-3 n-2 n-1 n n+1 n+2 n+3 n+4 Z, HI 6860 Los Alamos PAL Scaling Characteristics CCS-3 256PEs 7 64PEs 1 4 8PEs 3 8 2 5 2PEs 1 5 2 ... ... 9 12345678 4 6 12 3 3 2 4 7 Comparison of Boundary Surfaces Comparison of inter-PE communication distances 1.E+06 100 Neighbor Distance Grid Surface PE Surface 1.E+05 10 1.E+04 PE Distance Total Surface (cells) i) Surface size ii) PE Distance 1.E+03 1 110100 1000 10000 100000 110100 1000 10000 100000 # PEs # PEs Surface split across PEs: P >÷(E/8) PE distance = ÈÈ (8P2/E)1/3 ˘˘ (for cells/PE = 13,500 P > 41) Los Alamos PAL Validation CCS-3 G Validated on large-scale platforms: SAGE Performance (ASCI White) 2 – ASCI Blue Mountain (SGI Origin 2000) 1.8 1.6 – CRAY T3E 1.4 – ASCI Red (intel) 1.2 1 – ASCI White (IBM SP3) 0.8 Cycle Time (s) 0.6 – Compaq Alphaserver SMP clusters 0.4 Prediction 0.2 Measurement G 0 Model is highly accurate (< 10% error) 110100 1000 10000 # PEs SAGE Performance (ASCI Blue Mountain) SAGE Performance (Compaq ES45 AlphaServer) 6 0.6 5 0.5 4 0.4 3 0.3 Cycle Time (s) Cycle Time (s) 2 0.2 1 Prediction 0.1 Prediction Measurement Measurement 0 0 100 1000 10000 110100 # PEs # PEs Los Alamos PAL Experience: Compaq Installation CCS-3 G Model provided expected performance G Installation performed in stages Late 2001: Early 2002: upgraded PCI SAGE - Performance Data (Compaq ES45) SAGE - Performance Data (Compaq ES45) 1.6 ModelModel 1.6 Model ModelModel MeasuredMeasured (Sept(Sept 9th9th 01)01) 4-Jan4-Jan Measured (Oct 2nd 01) 2-Feb2-Feb 1.2 Measured (Oct 24th 01) 1.2 20-Apr 0.8 0.8 Cycle Time (s) Cycle Time (s) Cycle Time (s) Cycle Time (s) Cycle Time (s) Cycle Time (s)Cycle Time (s)Cycle Time (s) 0.4 0.4 0 0 110100 1000 110100 1000 10000 # PEs # PEs -> Model used to validate measurements! Other factors: accurately predicted when 2-rails improves the performance (P>41) Los Alamos PAL More recently: (Jan 2003) CCS-3 SAGE Performance (QA & QB) 1.2 1 0.8 0.6 Cycle-time (s)Cycle-time (s)Cycle-time (s)Cycle-time (s)Cycle-time (s) 0.4 ModelModel Sep-21-02 0.2 Nov-25-02 Jan-25-03 Jan-27-03 0 0 512 1024 1536 2048 2560 3072 3584 4096 # PEs G Performance of ASCI Q is now with ~10% of our expectation G Without a model we would not have identified (and solved) the poor performance! Los Alamos PAL SWEEP3D Particle Transport: 2-D CCS-3 Pipeline Parallelism n+3 n+2 n+1 n WW n-1 PE n-2 n-3 G SN transport using discrete ordinates method G wavefronts of computation propagate across grid G Model for Sweep3D accurately captures the performance Los Alamos PAL CCS-3 How many _ ? G Basis for Performance Comparison – Use validated performance models (trusted) to predict performance – Large problem size (Weak scaling) to fill available memory -> Problem size on Q is double that on the Earth Simulator G Compare processing rate: – Equal processor count basis (1 to 5120) – % of system used (10 to 100%) G But have an unknown: – performance of codes on single Earth Simulator processor -> Consider range of values Los Alamos PAL Model Parameters: SAGE CCS-3 Parameter Alpha ES45 Earth Simulator PSMP Processors per node 4 8 CL Communication Links per Node 1 1 E Number of level 0 cells 35,000 17,500 fGS_r Frequency of real and integer 377 377 fGS_I gather-scatters per cycle 22 22 T (E) Sequential Computation time per 68.6 s Ï 42.9m s (%5 of peak) comp Ô cell Ô 21.4m s (%10 of peak) Ô 14.3m s (%15 of peak) Ì Ô 10.7m s (%20 of peak) Ô 8.6m s (%25 of peak) Ô ÓÔ 7.1m s (%30 of peak) L (S,P) Bi-directional MPI communication Ï 6.10m s S < 64 8 s c Ô Latency Ì 6.44m s 64 £ S £ 512 Ô Ó 13.8m s S > 512 B (S,P) Bi-directional MPI communication Ï 0.0 S < 64 10GB/s c Ô Bandwidth (per direction) Ì 78MB/ s 64 £ S £ 512 Ô Ó 120MB/ s S > 512 Los Alamos PAL SAGE: Processing Rate CCS-3 16000 ASCI Q 14000 ASCI White 12000 10000 8000 Higher is better 6000 4000 Processing Rate (Cell-Cycles/s/PE) 2000 0 110100 1000 10000 #Processors G Weak-scaling G ASCI Q between 2.2 and 2.8 times faster than ASCI White G Data from Model (validated to 4096 PEs on Q, 8192 PEs on White) Los Alamos PAL SAGE: Processing Rate CCS-3 160000 30% 25% 140000 20% 15% Earth Simulator 120000 10% 5% 100000 80000 Higher is better 60000 40000 Processing Rate (Cell-Cycles/s/PE) 20000 0 110100 1000 10000 #Processors G Single Processor time is unknown – modeled using range of values Los Alamos PAL SAGE: Earth Simulator vs Q CCS-3 (Equal processor count) 12 11 10 9 8 7 Higher = Earth Simulator 6 Faster ASCI Q) 5 4 3 2 30% 25% 20% Relative Performance (Earth Simulator / 1 15% 10% 5% 0 110100 1000 10000 #Processors G Ratio of processor peak-performance is 3.2 (red line) G In nearly all cases: Earth Simulator performance is better than ratio of processor peak-performance Los Alamos PAL SAGE: Earth Simulator vs Q CCS-3 (% of system utilized) 8 7 6 5 Higher = Earth Simulator 4 Faster ASCI Q) 3 2 1 30% 25% 20% Relative Performance (Earth Simulator / 15% 10% 5% 0 0102030405060708090100 System Size (%) G Ratio of system peak-performance is 2 (red line) G In nearly all cases: Earth Simulator better than ratio of system peak-performance Los Alamos CCS-3 PAL SAGE: Component Time Predictions 100% 100% Compute/memory Compute/memory 90% 90% Bandwidth Bandwidth 80% Latency
Load more

Recommended publications
  • Project Aurora 2017 Vector Inheritance
    Vector Coding a simple user’s perspective Rudolf Fischer NEC Deutschland GmbH Düsseldorf, Germany SIMD vs. vector Input Pipeline Result Scalar SIMD people call it “vector”! Vector SX 2 © NEC Corporation 2017 Data Parallelism ▌‘Vector Loop’, data parallel do i = 1, n Real, dimension(n): a,b,c a(i) = b(i) + c(i) … end do a = b + c ▌‘Scalar Loop’, not data parallel, ex. linear recursion do i = 2, n a(i) = a(i-1) + b(i) end do ▌Reduction? do i = 1, n, VL do i_ = i, min(n,i+VL-1) do i = 1, n s_(i_) = s_(i_) + v(i_)* w(i_) s = s + v(i)* w(i) end do end do end do s = reduction(s_) (hardware!) 3 © NEC Corporation 2017 Vector Coding Paradigm ▌‘Scalar’ thinking / coding: There is a (grid-point,particle,equation,element), what am I going to do with it? ▌‘Vector’ thinking / coding: There is a certain action or operation, to which (grid-points,particles,equations,elements) am I going to apply it simultaneously? 4 © NEC Corporation 2017 Identifying the data-parallel structure ▌ Very simple case: do j = 2, m-1 do i = 2, n-1 rho_new(i,j) = rho_old(i,j) + dt * something_with_fluxes(i+/-1,j+/-1) end do end do ▌ Does it tell us something? Partial differential equations Local theories Vectorization is very “natural” 5 © NEC Corporation 2017 Identifying the data-parallel subset ▌ Simple case: V------> do j = 1, m |+-----> do i = 2, n || a(i,j) = a(i-1,j) + b(i,j) |+----- end do V------ end do ▌ The compiler will vectorise along j ▌ non-unit-stride access, suboptimal ▌ Results for a certain case: Totally scalar (directives): 25.516ms Avoid outer vector loop:
    [Show full text]
  • 2020 Global High-Performance Computing Product Leadership Award
    2020 GLOBAL HIGH-PERFORMANCE COMPUTING PRODUCT LEADERSHIP AWARD Strategic Imperatives Frost & Sullivan identifies three key strategic imperatives that impact the automation industry: internal challenges, disruptive technologies, and innovative business models. Every company that is competing in the automation space is obligated to address these imperatives proactively; failing to do so will almost certainly lead to stagnation or decline. Successful companies overcome the challenges posed by these imperatives and leverage them to drive innovation and growth. Frost & Sullivan’s recognition of NEC is a reflection of how well it is performing against the backdrop of these imperatives. Best Practices Criteria for World-Class Performance Frost & Sullivan applies a rigorous analytical process to evaluate multiple nominees for each award category before determining the final award recipient. The process involves a detailed evaluation of best practices criteria across two dimensions for each nominated company. NEC excels in many of the criteria in the high-performance computing (HPC) space. About NEC Established in 1899, NEC is a global IT, network, and infrastructure solution provider with a comprehensive product portfolio across computing, data storage, embedded systems, integrated IT infrastructure, network products, software, and unified communications. Headquartered in Tokyo, Japan, NEC has been at the forefront of accelerating the industrial revolution of the 20th and 21st © Frost & Sullivan 2021 The Growth Pipeline Company™ centuries by leveraging its technical knowhow and product expertise across thirteen different industries1 in industrial and energy markets. Deeply committed to the vision of orchestrating a better world. NEC envisions a future that embodies the values of safety, security, fairness, and efficiency, thus creating long-lasting social value.
    [Show full text]
  • Hardware Technology of the Earth Simulator 1
    Special Issue on High Performance Computing 27 Architecture and Hardware for HPC 1 Hardware Technology of the Earth Simulator 1 By Jun INASAKA,* Rikikazu IKEDA,* Kazuhiko UMEZAWA,* 5 Ko YOSHIKAWA,* Shitaka YAMADA† and Shigemune KITAWAKI‡ 5 1-1 This paper describes the hardware technologies of the supercomputer system “The Earth Simula- 1-2 ABSTRACT tor,” which has the highest performance in the world and was developed by ESRDC (the Earth 1-3 & 2-1 Simulator Research and Development Center)/NEC. The Earth Simulator has adopted NEC’s leading edge 2-2 10 technologies such as most advanced device technology and process technology to develop a high-speed and high-10 2-3 & 3-1 integrated LSI resulting in a one-chip vector processor. By combining this LSI technology with various advanced hardware technologies including high-density packaging, high-efficiency cooling technology against high heat dissipation and high-density cabling technology for the internal node shared memory system, ESRDC/ NEC has been successful in implementing the supercomputer system with its Linpack benchmark performance 15 of 35.86TFLOPS, which is the world’s highest performance (http://www.top500.org/ : the 20th TOP500 list of the15 world’s fastest supercomputers). KEYWORDS The Earth Simulator, Supercomputer, CMOS, LSI, Memory, Packaging, Build-up Printed Wiring Board (PWB), Connector, Cooling, Cable, Power supply 20 20 1. INTRODUCTION (Main Memory Unit) package are all interconnected with the fine coaxial cables to minimize the distances The Earth Simulator has adopted NEC’s most ad- between them and maximize the system packaging vanced CMOS technologies to integrate vector and density corresponding to the high-performance sys- 25 25 parallel processing functions for realizing a super- tem.
    [Show full text]
  • Supercomputers – Prestige Objects Or Crucial Tools for Science and Industry?
    Supercomputers – Prestige Objects or Crucial Tools for Science and Industry? Hans W. Meuer a 1, Horst Gietl b 2 a University of Mannheim & Prometeus GmbH, 68131 Mannheim, Germany; b Prometeus GmbH, 81245 Munich, Germany; This paper is the revised and extended version of the Lorraine King Memorial Lecture Hans Werner Meuer was invited by Lord Laird of Artigarvan to give at the House of Lords, London, on April 18, 2012. Keywords: TOP500, High Performance Computing, HPC, Supercomputing, HPC Technology, Supercomputer Market, Supercomputer Architecture, Supercomputer Applications, Supercomputer Technology, Supercomputer Performance, Supercomputer Future. 1 e-mail: [email protected] 2 e-mail: [email protected] 1 Content 1 Introduction ..................................................................................................................................... 3 2 The TOP500 Supercomputer Project ............................................................................................... 3 2.1 The LINPACK Benchmark ......................................................................................................... 4 2.2 TOP500 Authors ...................................................................................................................... 4 2.3 The 39th TOP500 List since 1993 .............................................................................................. 5 2.4 The 39th TOP10 List since 1993 ...............................................................................................
    [Show full text]
  • Beyond Earth Simulator
    Beyond Earth Simulator International Computing for the Atmospheric Sciences Symposium (iCAS2017) 海洋研究開発機構 September 11, 2017 地球情報基盤センターMakoto Tsukakoshi Information 情報システム部Systems Department Center for Earth Information Science and塚越 Technology眞 JAMSTEC Japan Agency for Marine-Earth Science and Technology 2 The main seven research and development issues during the third mid-term plan During the third mid-term plan, we set and address the seven research and development issues with all our strength due to promote strategic and focused research and development based on the national and social needs. Exploring untapped Ocean drilling – submarine resources Getting to know the Earth from beneath the seabed Detecting signals of global Information Science - environmental Predicting the Earth's change future by simulations Understanding seismogenic zones, and contributing to disaster mitigation Marine Bioscience - Exploring the unknown Construction of research base extreme biosphere to solve to spawn the ocean frontier the mystery of life 3 Typhoon-marine interaction using nonstatic atmospheric wave ocean coupling model Typhoon Vera in 1959 by Kazuhisa Tsuboki (2015) Typhoon-Ocean Interaction Study Using the Coupled Atmosphere-Ocean Non-hydrostatic Model: With Careful Consideration of Upper Outflow Layer Clouds of Typhoon 4 • Earth Simulator ： - Developed by NASDA, JAERI and JAMSTEC (563.4* Oku Yen) - Operation started on March 2002 and immediately recognized as #1 supercomputer - “It will keep #1 for 2 years in peak and 5 years in effective performance” by H. Miyoshi *inclusing facilities and building 5 Earth Simulator has changed its role when “K computer” is completed Council for Science and Technology Policy (CSTP) Report on Nov 25, 2005: After the completion of Leading Edge High Performance Supercomputer (i.e.
    [Show full text]
  • Evaluation of Cache-Based Superscalar and Cacheless Vector Architectures for Scientific Computations
    Evaluation of Cache-based Superscalar and Cacheless Vector Architectures for Scientific Computations Leonid Oliker, Jonathan Carter, John Shalf, David Skinner CRDLVERSC, Lawrence Berkeley National Laboratory, Berkeley, CA 94720 Stephane Ethier Princeton Plasma Physics Laboratory, Princeton Universiq, Princeton, NJ 08453 Rupak Biswas, Jahed Djomehri*, and Rob Van der Wijngaart* NAS Division, NASA Ames Research Centec Mofett Field, CA 94035 Abstract The growing gap between sustained and peak performance for scientific applications has become a well-known problem in high performance computing. The recent development of parallel vector systems offers the potential to bridge this gap for a significant number of computational science codes and deliver a substantial increase in computing capabilities. This paper examines the intranode performance of the NEC SX6 vector processor and the cache-based IBM Power3/4 superscalar architectures across a number of key scientific computing areas. First, we present the performance of a microbenchmark suite that examines a full spectrum of low-level machine characteristics. Next, we study the behavior of the NAS Parallel Benchmarks using some simple optimizations. Finally, we evaluate the perfor- mance of several numerical codes from key scientific computing domains. Overall results demonstrate that the SX6 achieves high performance on a large fraction of our application suite and in many cases significantly outperforms the RISC-based architectures. However, certain classes of applications are not easily amenable to vectorization and would likely require extensive reengineering of both algorithm and implementation to utilize the SX6 effectively. 1 Introduction The rapidly increasing peak performance and generality of superscalar cache-based microprocessors long led re- searchers to believe that vector architectures hold little promise for future large-scale computing systems.
    [Show full text]
  • Japan's 10 Peta FLOPS Supercomputer Development
    Japan’s 10 Peta FLOPS Supercomputer Development Project and its energy saving designs Ryutaro Himeno 1) Group Director of R&D 2) Deputy Program Director 3) Director of Advance group, NGS R&D Center for Computational Science Center for Computer & Communication Contents 1. Self introduction 2. Japan’s next generation supercomputer R&D project: 1. Before the project started 2. Introduction and current status 3. Energy issues 3. Introduction of focused application area 2 Who am I? 1985 1987 1988 1995 1993 1990 Courtesy of Nissan 1992 3 4 Hardware development at RIKEN Special purpose computers MD-GRAPE2 Theoretical Peak In 1989, GRAPE Performance developed by U. Tokyo 100Peta MD-GRAPE3 MDM:MD-GRAPE2 develop by RIKEN 10Peta Japan US GRAPE-DR MDM was integrated in (U.Tokyo/RIKEN) BlueGene/Q installed 1PetaFLOPS RSCC, 2004 MD-GRAPE3 1Peta Planed HPCS MD-GRAPE3 developed in (RIKEN) NLCF BlueGene/L 2004 367TeraFLOPS 100 ASC nd Purple 2 fastest record on MD- Tera Earth Simulator ASC Tsubame 41TeraFLOPS Red (TITech) GRAPE3: Gordon Bell Storm ASCI Q PACS-CS Prize, 2006. 10Tera RSCC ASCI (Tsukuba) Blue Pacific ASCI White (Riken) MD-GRAPE-3 will be ASCI Red integrated in RSCC soon. ASCI Blue Mountain 1Tera *ASCI: Accelerated Strategic CP-PACS Computing Initiative '96 '98 '00 '02 '04 '06 '08 '10 [year] 5 Current RSCC System installed in 2004 at Riken Advance Center for Computing & Communication HDD User User Tape Storage User 20 TB User 200 TB POWER4 Server: 7 Itanium2 Server: 2 Front end Tape Drive: 4 (16) Server Disk Cache: 2TB Myrinet InfiniBand
    [Show full text]
  • 1. Introduction
    Recent Trends in the Marketplace of High Performance Computing Erich Strohmaier1, Jack J. Dongarra2, Hans W. Meuer3, and Horst D. Simon4 High Performance Computing, HPC Market, Supercomputer Market, HPC technology, Supercomputer market, Supercomputer technology In this paper we analyze major recent trends and changes in the High Performance Computing (HPC) market place. The introduction of vector computers started the area of ‘Supercomputing’. The initial success of vector computers in the seventies was driven by raw performance. Massive Parallel Systems (MPP) became successful in the early nineties due to their better price/performance ratios, which was enabled by the attack of the ‘killer-micros’. The success of microprocessor based SMP concepts even for the very high-end systems, was the basis for the emerging cluster concepts in the early 2000s. Within the first half of this decade clusters of PC’s and workstations have become the prevalent architecture for many HPC application areas on all ranges of performance. However, the Earth Simulator vector system demonstrated that many scientific applications could benefit greatly from other computer architectures. At the same time there is renewed broad interest in the scientific HPC community for new hardware architectures and new programming paradigms. The IBM BlueGene/L system is one early example of a shifting design focus for large-scale system. The DARPA HPCS program has the declared goal of building a Petaflops computer system by the end of the decade using novel computer architectures.
    [Show full text]
  • Notes on the Earth Simulator
    Notes on the Earth Simulator Jack Dongarra Computer Science Department University of Tennessee FASTEST COMPUTER TODAY The Japanese Earth Simulator Research and Development Center Japan Atomic Energy Research Institute Atmospheric and Solid earth science oceanographic science High resolution Global dynamic model global models predictions of to describe the entire solid global warming etc earth as a system. High resolution Earth Earth Regional model regional models Simulator predictions of to describe crust/mantle El Niño events and activity in the Japanese Asian monsoon etc., Archipelago region, High resolution local models Simulation of earthquake predictions of weather disasters generation process such as typhoons, localized Seismic wave tomography torrential downpour, oil spill, downburst etc. 1 Earth Simulator • Based on the NEC SX architecture, 640 nodes, each node with 8 vector processors (8 Gflop/s peak per processor), 2 ns cycle time, 16GB shared memory. – Total of 5104 total processors, 40 TFlop/s peak, and 10 TB memory. • It has a single stage crossbar (1800 miles of cable) 83,000 copper cables, 16 GB/s cross section bandwidth. • 700 TB disk space • 1.6 PB mass store • Area of computer = 4 tennis courts, 3 floors Outline of the Earth Simulator Computer • Architecture : A MIMD-type, distributed memory, parallel system consisting of computing nodes in which vector-type multi- processors are tightly connected by sharing main memory. • Total number of processor nodes: 640 • Number of PE’s for each node: 8 • Total number of PE’s: 5120 • Peak performance of each PE: 8 GFLOPS • Peak performance of each node: 64 GFLOPS • Main memory : 10 TB (total).
    [Show full text]
  • The TOP500 Project Looking Back Over 16 Years of Supercomputing Experience with Special Emphasis on the Industrial Segment
    The TOP500 Project Looking Back Over 16 Years of Supercomputing Experience with Special Emphasis on the Industrial Segment Hans Werner Meuer [email protected] Prometeus GmbH & Universität Mannheim Microsoft Industriekunden – Veranstaltung Frankfurt /16. Oktober 2008 / Windows HPC Server 2008 Launch page 1 31th List: The TOP10 Rmax Power Manufacturer Computer Installation Site Country #Cores [TF/s] [MW] Roadrunner 1 IBM 1026 DOE/NNSA/LANL USA 2.35 122,400 BladeCenter QS22/LS21 BlueGene/L 2 IBM 478.2 DOE/NNSA/LLNL USA 2.33 212,992 eServer Blue Gene Solution Intrepid 3 IBM 450.3 DOE/ANL USA 1.26 163,840 Blue Gene/P Solution Ranger 4 Sun 326 TACC USA 2.00 62,976 SunBlade x6420 Jaguar 5 Cray 205 DOE/ORNL USA 1.58 30,976 Cray XT4 QuadCore JUGENE Forschungszentrum 6 IBM 180 Germany 0.50 65,536 Blue Gene/P Solution Juelich (FZJ) Encanto New Mexico Computing 7 SGI 133.2 USA 0.86 14,336 SGI Altix ICE 8200 Applications Center EKA Computational Research 8 HP 132.8 India 1.60 14,384 Cluster Platform 3000 BL460c Laboratories, TATA SONS 9 IBM Blue Gene/P Solution 112.5 IDRIS France 0.32 40,960 Total Exploration 10 SGI SGI Altix ICE 8200EX 106.1 France 0.44 10,240 Production Microsoft Industriekunden – Veranstaltung page 2 Outline Mannheim Supercomputer Statistics & Top500 Project Start in 1993 Competition between Manufacturers, Countries and Sites My Supercomputer Favorite in the Top500 Lists The 31st List as of June 2008 Performance Development and Projection Bell‘s Law Supercomputing, quo vadis? Top500, quo vadis? Microsoft Industriekunden – Veranstaltung
    [Show full text]
  • Porting the 3D Gyrokinetic Particle-In-Cell Code GTC to the NEC SX-6 Vector Architecture: Perspectives and Challenges
    Computer Physics Communications 164 (2004) 456–458 www.elsevier.com/locate/cpc Porting the 3D gyrokinetic particle-in-cell code GTC to the NEC SX-6 vector architecture: perspectives and challenges S. Ethier a,∗,Z.Linb a Princeton Plasma Physics Laboratory, Princeton, NJ 08543, USA b Department of Physics and Astronomy, University of California, Irvine, CA 92697, USA Available online 27 October 2004 Abstract Several years of optimization on the cache-based super-scalar architecture has made it more difficult to port the current version of the 3D particle-in-cell code GTC to the NEC SX-6 vector architecture. This paper explains the initial work that has been done to port this code to the SX-6 computer and to optimize the most time consuming parts. After a few modifications, single-processor results show a performance increase of 5.2 compared to the IBM SP Power3 processor, and 2.7 compared to the Power4. 2004 Elsevier B.V. All rights reserved. PACS: 02.70.Ns; 07.05.Tp Keywords: Vector processor; Particle-in-cell; Code optimization 1. Introduction (< 10%). When properly vectorized, the same codes can, however, reach over 30 or even 40% of peak per- The impressive performance achieved in 2002 by formance on a vector processor. Not all codes can the Japanese Earth Simulator computer (26.58 Tflops, achieve such performance. The purpose of this study 64.9% of peak) [1] has revived the interest in vec- is to evaluate the work/reward ratio involved in vec- tor processors. Having been the flagship of high per- torizing our particle-in-cell code on the latest paral- formance computing for more than two decades, vec- lel vector machines, such as the NEC SX-6, which is tor computers were gradually replaced, at least in the building block of the very large Earth Simulator the US, by much cheaper multi-processor super-scalar system in Japan [2].
    [Show full text]
  • SX-Aurora TSUBASA Introduction Vector Supercomputer Technology on a Pcie Card What Is Vector Processor? (1/2)
    SX-Aurora TSUBASA Introduction Vector Supercomputer Technology on a PCIe Card What is Vector Processor? (1/2) Vector processor can operate large data at once and suited for fast processing of large scale data General Processor Vector Processor Suited for processing data in small Suited for processing data in large units such as business operation and units at once such as simulation,AI, web servers and Bigdata data data 256 Scalar Vector calculation calculation 256 output output 2 © NEC Corporation 2019 What is Vector Processor? (2/2) ① Many small cores vs small number of large cores ② Balance of computation performance and data access performance ③ Software development environment GPU-like Processors Vector Processors ① Many small cores ① Small number of large cores ② Larger size of computation circuits ② Balanced size of computation circuits and ③ Special language (such as CUDA) data access circuits ③ Standard language (C/C++/Fortran) Cores Cores Data access Data access Memory Memory 3 © NEC Corporation 2018 Vector Processor – History & Future ▌Vector Processor has traditionally been used to process big data, much earlier than the term big data was coined. ▌The very first vector processor based machine, Cray-1, was built by Seymour Cray in 1976. NEC made its first vector-supercomputer, the SX-2, in 1981. SX-2 was the first ever CPU to exceed 1 Gflops of peaK performance. Soon, Fujitsu, Hitachi followed NEC’s footsteps in the high-end HPC Technology segment. ▌However, in 1990s, the computer industry changed drastically with the advent of affordable x86 processors. The eventual dominance of x86 played a Key-role in democratization of HPC across academia & industry.
    [Show full text]