Scalable Single System Image SGI Altix 3700, 512p Architecture and Software Environment
|
|
|
- Geraldine Higgins
- 5 years ago
- Views:
Transcription
1 Silicon Graphics, Inc. Scalable Single System Image SGI Altix 3700, 512p Architecture and Software Environment Presented by: Jean-Pierre Panziera Principal Engineer
2 Altix 3700 SSSI - Architecture and Software Environment SP and NUA Directory based Shared emory SGI Altix 3700, scaling Hardware to 512p Scaling Linux to 512p Scaling Applications to 512p next
3 Symmetric ultiprocessor SP CPU CPU CPU CPU Interconnection CPU CPU CPU CPU Bus E E E E CPU CPU CPU CPU E E E E Switch E E E E
4 Non Uniform emory Access NUA node 0 node 1 node 2 node 3 node n CPU CPU CPU CPU CPU E E E E E Interconnection
5 Sharing data on a Cache Coherent NUA system ccnua Node m eg L2 L3 FSB em Node k Node n eg L2 L3 Interconnection Fabric FSB
6 ESI protocol Cache Coherence on the CPU side Node n eg L2 L3 FSB For each cacheline (128Bytes) coherency state is either: odified Dirty Exclusive DEX E xclusive Clean Exclusive CEX S hared Shared SHD I nvalid Invalid INV
7 Directory based coherency Cache Coherence on the emory side Node k em dir Directory information is in memory DIs Directory takes only 3% on total memory For each CacheLine available information : Line state (Unowned,Exclusive,Shared,Busy, ) 2 bits Sharing Vector, list of nodes (256 max) 24 bits Priority Protection ECC
8 Directory based Cache Coherence sequences of sharing messages 2 N 3 K NodeN CEX BL(shared) INTE DNGD emory N EXCL 1 0 NodeK EAD ESPEC SACK INV INV Node load SHD SHD 1 1 SHD N K INV load INV EAD ESPEC EPLY... EXCL 0 1 BUSY 0 0 INTE XFE... DEX store BL(shared) INV CEX EXCL 1 0
9 SGI Altix 3700 Hardware CPU-emory brick SC-Brick (3 U) Node 0 Node 1 emory emory Itanium 2 FSB Itanium 2 SHub emory 10.2GB/s NL4 6.4GB/s SHub 2x16 GB Itanium GB/s Itanium 2 Numalink T 3/4 3.2 / 6.4 GB/s XIO T 2.4 GB/s
10 SGI Altix Processors in a ack NUAlink 3 outers 2x1.6=3.2GB/s Bisection BW 12.8 GB/s 400 B/s/p
11 Altix 3700 fat-tree topology for 512p/4TB system Bisection BW GB/s 400 B/s/p Image courtesy: NASA Ames
12 Linux for Altix 3000 / Linux 2.4 kernel SGI Value-Added Enhancements SGI Open Source Enhancements Standard Linux Distribution Boot/Driver CD Differentiating Features and Functionality CPU sets / emory placement PT & Array Services Hierarchical storage management tools Partitioning XV, XSCSI PCP Enabling Features and Functionality Latest bugfixes, other supported device drivers, etc Comprehensive system accounting (CSA) Job containers (PAGG) XFS Base OS and Common Open Source Apps Platform support, error handling, scaling, NUA O(1) Scheduler Device Drivers, SGI XFS installer SCSL CHASE CXFS dplace runon
13 Linux for Altix 3000 NUA support discontiguous memory support V Support text replication process to processor binding, local memory allocation Partitioning support (shared memory clusters) e.g. configure 256p as 4x64p Scalability enhancements Fixes
14 O(1) Scheduler Scheduler improvements required Standard Linux scheduler poor cache usage; too many task migrations heavy contention for runqueue_lock O(1) scheduler tasks stick to processors no global runqueue_lock (multiple run queues) 6x improvement on some benchmarks Still needs work migration livelock with lots of idle processors (fixed) NUA awareness (being worked by us & community) Fairness in overcomitted workload situations no gang scheduling concept
15 CPU and emory Allocation equired to get good NUA performance Want memory allocated locally if possible exploit local memory bandwidth Want to bind processes to specific processors to avoid cache damage caused by migration to keep processes close to where data allocated Want processors working on single HPC application to be close to one another to minimize memory latency when cross-processor memory references are required Don t want to have to recompile the application
16 CPU and emory Allocation dplace command: dplace -c0-7 -s3 program assumes 9 processes will be created 8 will be bound to CPUs 0-7 3rd process created will be skipped not bound, e. g. for pthreads shepherd processes Storage will be allocated locally first-touch allocation rule cpusets provides finer grain control, e.g. : boot cpuset for interactive jobs and kernel data compute cpuset(s) to isolate applications
17 Stream Triad Benchmark esults (GB/s) SGI Altix 3000 (512P) NEC SX-7 (32P) NEC SX-5 (16P) SGI Altix 3000 (256P) NEC SX-4 (32P) HP AlphaServer GS1280 (64P) Cray T3E (32P) SGI Altix 3000 (128P) NEC SX-6 (8P) SGI Altix 3000 (64P) Cray C90 (16P) SGI Origin 3800 (256P) HP Integrity SuperDome (64P) SGI Altix 3000 (32P) IB eserver p690+ (32P) Sun Fire15K (72P) SGI Origin 2000 (256p) Cray SV1 (32P) HP AlphaServer ES80 (8P) IB eserver p670+ (16P) IB eserver p690 Turbo (32P) SGI Altix 3000 (16P) Cray Y-P (8P) HP SuperDome 750 (64P) IB eserver p690 HPC (16P) 127,3 103,8 99,7 84,2 63,7 58,9 50,7 49,3 47,8 44,5 36,8 32,2 31,9 26,8 26,5 25, ,7 359,3 488,3 583,1 872,3 1007,83 easures memory bandwidth performance Simple loops Embarrassingly parallel Easy for compiler to generate scalable code No fancy flags: -O3 STEA Triad:!$OP PAALLEL DO DO j = 1,n a(j)=b(j)+s*c(j) CONTINUE Source: SGI Altix 3000 Series Performance eport 1.6 (9/10/03), Customer provided data, & STEA website (11/13/03)
18 PI across partitions: Latency vs. Distance PI Point-to-Point Latency (from CPU 0) 2x64p 1.5 GHz Altix 3700 Supercluster Using PT 1.10 on ProPack 2.4 2,5 8 byte transfer size 2 Time (usec) 1,5 1 0,5 Host A Host B Destination CPU
19 IFS (ECWF) a scalable PI application Total Time Communication Time Parallel Speedup ideal speedup Time efficiency 78% Speedup Number of processors
20 Scaling OpenP Cart3D (Aerodynamics) on SGI Altix at NASA Ames CPUs Achieved % efficiency CPUs Used Source: NASA Ames esearch Center
21 Efficient Shared emory Parallelism SP Pthreads / OpenP Highest possible level parallelism (forget incremental) Limit global commons/arrays to necessary ake sharing explicit Use [thread]private local variables/arrays eplicate global data to local space emove unnecessary synchronization emove big global locks individual locks for critical code or data SP enables Dynamic Load Balancing
22 SGI Altix 3000 future evolutions Faster Itanium 2 processors NUAlink 4 routers (2xNL3 BW) Higher processor density educed emote Latency Faster/Denser emory DIs Faster FSB (front side bus) Larger Cache Coherence domains Larger Shared emory systems
23 Next generation: UV Architecture Vision PU PU IO APU UV Petascale GA. Globally Addressable emory. Low Latency. High Bandwidth. O(10K) Ports IO APU (GPU) APU APU APU (GPU)
24 Altix 3700 SSSI - Architecture and Software environment Questions?
SMP and ccnuma Multiprocessor Systems. Sharing of Resources in Parallel and Distributed Computing Systems
Reference Papers on SMP/NUMA Systems: EE 657, Lecture 5 September 14, 2007 SMP and ccnuma Multiprocessor Systems Professor Kai Hwang USC Internet and Grid Computing Laboratory Email: kaihwang@usc.edu [1]
COMP Parallel Computing. CC-NUMA (1) CC-NUMA implementation
COP 633 - Parallel Computing Lecture 10 September 27, 2018 CC-NUA (1) CC-NUA implementation Reading for next time emory consistency models tutorial (sections 1-6, pp 1-17) COP 633 - Prins CC-NUA (1) Topics
HPCS HPCchallenge Benchmark Suite
HPCS HPCchallenge Benchmark Suite David Koester, Ph.D. () Jack Dongarra (UTK) Piotr Luszczek () 28 September 2004 Slide-1 Outline Brief DARPA HPCS Overview Architecture/Application Characterization Preliminary
LSF HPC :: getting most out of your NUMA machine
Leopold-Franzens-Universität Innsbruck ZID Zentraler Informatikdienst (ZID) LSF HPC :: getting most out of your NUMA machine platform computing conference, Michael Fink who we are & what we do university
Non-Uniform Memory Access (NUMA) Architecture and Multicomputers
Non-Uniform Memory Access (NUMA) Architecture and Multicomputers Parallel and Distributed Computing Department of Computer Science and Engineering (DEI) Instituto Superior Técnico February 29, 2016 CPD
1. Memory technology & Hierarchy
1. Memory technology & Hierarchy Back to caching... Advances in Computer Architecture Andy D. Pimentel Caches in a multi-processor context Dealing with concurrent updates Multiprocessor architecture In
Non-Uniform Memory Access (NUMA) Architecture and Multicomputers
Non-Uniform Memory Access (NUMA) Architecture and Multicomputers Parallel and Distributed Computing Department of Computer Science and Engineering (DEI) Instituto Superior Técnico September 26, 2011 CPD
Requirements for Scalable Application Specific Processing in Commercial HPEC
Requirements for Scalable Application Specific Processing in Commercial HPEC Steven Miller Silicon Graphics, Inc. Phone: 650-933-1899 Email Address: scm@sgi.com Abstract: More and more High Performance
Considerations for LS-DYNA Workflow Efficiencies in an HPC Linux Environment
9 th International LS-DYNA Users Conference Computing / Code Technology (2) Considerations for LS-DYNA Workflow Efficiencies in an HPC Linux Environment Stanley Posey HPC Applications Development SGI,
Non-Uniform Memory Access (NUMA) Architecture and Multicomputers
Non-Uniform Memory Access (NUMA) Architecture and Multicomputers Parallel and Distributed Computing MSc in Information Systems and Computer Engineering DEA in Computational Engineering Department of Computer
COSC 6385 Computer Architecture - Multi Processor Systems
COSC 6385 Computer Architecture - Multi Processor Systems Fall 2006 Classification of Parallel Architectures Flynn s Taxonomy SISD: Single instruction single data Classical von Neumann architecture SIMD:
Parallel Computers. CPE 631 Session 20: Multiprocessors. Flynn s Tahonomy (1972) Why Multiprocessors?
Parallel Computers CPE 63 Session 20: Multiprocessors Department of Electrical and Computer Engineering University of Alabama in Huntsville Definition: A parallel computer is a collection of processing
Portland State University ECE 588/688. Directory-Based Cache Coherence Protocols
Portland State University ECE 588/688 Directory-Based Cache Coherence Protocols Copyright by Alaa Alameldeen and Haitham Akkary 2018 Why Directory Protocols? Snooping-based protocols may not scale All
Parallel Computing Platforms. Jinkyu Jeong Computer Systems Laboratory Sungkyunkwan University
Parallel Computing Platforms Jinkyu Jeong (jinkyu@skku.edu) Computer Systems Laboratory Sungkyunkwan University http://csl.skku.edu Elements of a Parallel Computer Hardware Multiple processors Multiple
Lecture 17. NUMA Architecture and Programming
Lecture 17 NUMA Architecture and Programming Announcements Extended office hours today until 6pm Weds after class? Partitioning and communication in Particle method project 2012 Scott B. Baden /CSE 260/
CS654 Advanced Computer Architecture Lec 14 Directory Based Multiprocessors
CS654 Advanced Computer Architecture Lec 14 Directory Based Multiprocessors Peter Kemper Adapted from the slides of EECS 252 by Prof. David Patterson Electrical Engineering and Computer Sciences University
Considerations for LS-DYNA Efficiency in SGI IRIX and Linux Environments with a NUMA System Architecture
4 th European LS-DYNA Users Conference MPP / Linux Cluster / Hardware I Considerations for LS-DYNA Efficiency in SGI IRIX and Linux Environments with a NUMA System Architecture Authors: Stan Posey, Nick
ECE 669 Parallel Computer Architecture
ECE 669 Parallel Computer Architecture Lecture 18 Scalable Parallel Caches Overview ost cache protocols are more complicated than two state Snooping not effective for network-based systems Consider three
Parallel Computing Platforms
Parallel Computing Platforms Jinkyu Jeong (jinkyu@skku.edu) Computer Systems Laboratory Sungkyunkwan University http://csl.skku.edu SSE3054: Multicore Systems, Spring 2017, Jinkyu Jeong (jinkyu@skku.edu)
Learning Curve for Parallel Applications. 500 Fastest Computers
Learning Curve for arallel Applications ABER molecular dynamics simulation program Starting point was vector code for Cray-1 145 FLO on Cray90, 406 for final version on 128-processor aragon, 891 on 128-processor
EE 4683/5683: COMPUTER ARCHITECTURE
3/3/205 EE 4683/5683: COMPUTER ARCHITECTURE Lecture 8: Interconnection Networks Avinash Kodi, kodi@ohio.edu Agenda 2 Interconnection Networks Performance Metrics Topology 3/3/205 IN Performance Metrics
Basic Architecture of SMP. Shared Memory Multiprocessors. Cache Coherency -- The Problem. Cache Coherency, The Goal.
Shared emory ultiprocessors Basic Architecture of SP Buses are good news and bad news The (memory) bus is a point all processors can see and thus be informed of what is happening A bus is serially used,
Getting Performance from OpenMP Programs on NUMA Architectures
Getting Performance from OpenMP Programs on NUMA Architectures Christian Terboven, RWTH Aachen University terboven@itc.rwth-aachen.de EU H2020 Centre of Excellence (CoE) 1 October 2015 31 March 2018 Grant
High-Performance Lustre with Maximum Data Assurance
High-Performance Lustre with Maximum Data Assurance Silicon Graphics International Corp. 900 North McCarthy Blvd. Milpitas, CA 95035 Disclaimer and Copyright Notice The information presented here is meant
Introduction to parallel computers and parallel programming. Introduction to parallel computersand parallel programming p. 1
Introduction to parallel computers and parallel programming Introduction to parallel computersand parallel programming p. 1 Content A quick overview of morden parallel hardware Parallelism within a chip
The Impact of Optics on HPC System Interconnects
The Impact of Optics on HPC System Interconnects Mike Parker and Steve Scott Hot Interconnects 2009 Manhattan, NYC Will cost-effective optics fundamentally change the landscape of networking? Yes. Changes
Birds of a Feather Presentation
Mellanox InfiniBand QDR 4Gb/s The Fabric of Choice for High Performance Computing Gilad Shainer, shainer@mellanox.com June 28 Birds of a Feather Presentation InfiniBand Technology Leadership Industry Standard
Convergence of Parallel Architecture
Parallel Computing Convergence of Parallel Architecture Hwansoo Han History Parallel architectures tied closely to programming models Divergent architectures, with no predictable pattern of growth Uncertainty
Computer Comparisons Using HPCC. Nathan Wichmann Benchmark Engineer
Computer Comparisons Using HPCC Nathan Wichmann Benchmark Engineer Outline Comparisons using HPCC HPCC test used Methods used to compare machines using HPCC Normalize scores Weighted averages Comparing
A Uniform Programming Model for Petascale Computing
A Uniform Programming Model for Petascale Computing Barbara Chapman University of Houston WPSE 2009, Tsukuba March 25, 2009 High Performance Computing and Tools Group http://www.cs.uh.edu/~hpctools Agenda
White Paper. Technical Advances in the SGI. UV Architecture
White Paper Technical Advances in the SGI UV Architecture TABLE OF CONTENTS 1. Introduction 1 2. The SGI UV Architecture 2 2.1. SGI UV Compute Blade 3 2.1.1. UV_Hub ASIC Functionality 4 2.1.1.1. Global
MULTIPROCESSORS AND THREAD-LEVEL. B649 Parallel Architectures and Programming
MULTIPROCESSORS AND THREAD-LEVEL PARALLELISM B649 Parallel Architectures and Programming Motivation behind Multiprocessors Limitations of ILP (as already discussed) Growing interest in servers and server-performance
MULTIPROCESSORS AND THREAD-LEVEL PARALLELISM. B649 Parallel Architectures and Programming
MULTIPROCESSORS AND THREAD-LEVEL PARALLELISM B649 Parallel Architectures and Programming Motivation behind Multiprocessors Limitations of ILP (as already discussed) Growing interest in servers and server-performance
The Multikernel: A new OS architecture for scalable multicore systems Baumann et al. Presentation: Mark Smith
The Multikernel: A new OS architecture for scalable multicore systems Baumann et al. Presentation: Mark Smith Review Introduction Optimizing the OS based on hardware Processor changes Shared Memory vs
Cache Coherence. Todd C. Mowry CS 740 November 10, Topics. The Cache Coherence Problem Snoopy Protocols Directory Protocols
Cache Coherence Todd C. Mowry CS 740 November 10, 1998 Topics The Cache Coherence roblem Snoopy rotocols Directory rotocols The Cache Coherence roblem Caches are critical to modern high-speed processors
Issues in Multiprocessors
Issues in Multiprocessors Which programming model for interprocessor communication shared memory regular loads & stores SPARCCenter, SGI Challenge, Cray T3D, Convex Exemplar, KSR-1&2, today s CMPs message
Intro to Multiprocessors
The Big Picture: Where are We Now? Intro to Multiprocessors Output Output Datapath Input Input Datapath [dapted from Computer Organization and Design, Patterson & Hennessy, 2005] Multiprocessor multiple
Introducing the Cray XMT. Petr Konecny May 4 th 2007
Introducing the Cray XMT Petr Konecny May 4 th 2007 Agenda Origins of the Cray XMT Cray XMT system architecture Cray XT infrastructure Cray Threadstorm processor Shared memory programming model Benefits/drawbacks/solutions
COSC4201 Multiprocessors
COSC4201 Multiprocessors Prof. Mokhtar Aboelaze Parts of these slides are taken from Notes by Prof. David Patterson (UCB) Multiprocessing We are dedicating all of our future product development to multicore
Multiprocessors II: CC-NUMA DSM. CC-NUMA for Large Systems
Multiprocessors II: CC-NUMA DSM DSM cache coherence the hardware stuff Today s topics: what happens when we lose snooping new issues: global vs. local cache line state enter the directory issues of increasing
Lecture 13: Memory Consistency. + a Course-So-Far Review. Parallel Computer Architecture and Programming CMU , Spring 2013
Lecture 13: Memory Consistency + a Course-So-Far Review Parallel Computer Architecture and Programming Today: what you should know Understand the motivation for relaxed consistency models Understand the
Parallel Architecture. Sathish Vadhiyar
Parallel Architecture Sathish Vadhiyar Motivations of Parallel Computing Faster execution times From days or months to hours or seconds E.g., climate modelling, bioinformatics Large amount of data dictate
Altix UV HW/SW! SGI Altix UV utilizes an array of advanced hardware and software feature to offload:!
Altix UV HW/SW! SGI Altix UV utilizes an array of advanced hardware and software feature to offload:!! thread synchronization!! data sharing!! massage passing overhead from CPUs.! This system has a rich
A Scalable SAS Machine
arallel omputer Organization and Design : Lecture 8 er Stenström. 2008, Sally. ckee 2009 Scalable ache oherence Design principles of scalable cache protocols Overview of design space (8.1) Basic operation
Silicon Graphics Prism TM : A Platform for Scalable Graphics
Silicon Graphics, Inc. Silicon Graphics Prism TM : A Platform for Scalable Graphics Presented by: Alpana Kaulgud Engineering Director, Visual Systems Bruno Stefanizzi Applications Engineering Silicon Graphics
Outline. Execution Environments for Parallel Applications. Supercomputers. Supercomputers
Outline Execution Environments for Parallel Applications Master CANS 2007/2008 Departament d Arquitectura de Computadors Universitat Politècnica de Catalunya Supercomputers OS abstractions Extended OS
Lecture 25: Multiprocessors. Today s topics: Snooping-based cache coherence protocol Directory-based cache coherence protocol Synchronization
Lecture 25: Multiprocessors Today s topics: Snooping-based cache coherence protocol Directory-based cache coherence protocol Synchronization 1 Snooping-Based Protocols Three states for a block: invalid,
Introduction to Multiprocessors (Part I) Prof. Cristina Silvano Politecnico di Milano
Introduction to Multiprocessors (Part I) Prof. Cristina Silvano Politecnico di Milano Outline Key issues to design multiprocessors Interconnection network Centralized shared-memory architectures Distributed
Network Based Coherency: Extending a Processor's Coherency Domain over a Standard Network
Network Based Coherency: Extending a Processor's Coherency omain over a Standard Network Bob Quinn, Isam kkawi, Krishnan Subramani, Shahe Krakirian ugust 2008 Why Network Based Coherency? Current data
2008 International ANSYS Conference
2008 International ANSYS Conference Maximizing Productivity With InfiniBand-Based Clusters Gilad Shainer Director of Technical Marketing Mellanox Technologies 2008 ANSYS, Inc. All rights reserved. 1 ANSYS,
Cray XE6 Performance Workshop
Cray XE6 erformance Workshop odern HC Architectures David Henty d.henty@epcc.ed.ac.uk ECC, University of Edinburgh Overview Components History Flynn s Taxonomy SID ID Classification via emory Distributed
Lecture 17: Parallel Architectures and Future Computer Architectures. Shared-Memory Multiprocessors
Lecture 17: arallel Architectures and Future Computer Architectures rof. Kunle Olukotun EE 282h Fall 98/99 1 Shared-emory ultiprocessors Several processors share one address space» conceptually a shared
NUMA replicated pagecache for Linux
NUMA replicated pagecache for Linux Nick Piggin SuSE Labs January 27, 2008 0-0 Talk outline I will cover the following areas: Give some NUMA background information Introduce some of Linux s NUMA optimisations
COEN-4730 Computer Architecture Lecture 08 Thread Level Parallelism and Coherence
1 COEN-4730 Computer Architecture Lecture 08 Thread Level Parallelism and Coherence Cristinel Ababei Dept. of Electrical and Computer Engineering Marquette University Credits: Slides adapted from presentations
Scalable Cache Coherence. Jinkyu Jeong Computer Systems Laboratory Sungkyunkwan University
Scalable Cache Coherence Jinkyu Jeong (jinkyu@skku.edu) Computer Systems Laboratory Sungkyunkwan University http://csl.skku.edu Hierarchical Cache Coherence Hierarchies in cache organization Multiple levels
COSC4201. Multiprocessors and Thread Level Parallelism. Prof. Mokhtar Aboelaze York University
COSC4201 Multiprocessors and Thread Level Parallelism Prof. Mokhtar Aboelaze York University COSC 4201 1 Introduction Why multiprocessor The turning away from the conventional organization came in the
The STREAM Benchmark. John D. McCalpin, Ph.D. IBM eserver Performance ^ Performance
The STREAM Benchmark John D. McCalpin, Ph.D. IBM eserver Performance 2005-01-27 History Scientific computing was largely based on the vector paradigm from the late 1970 s through the 1980 s E.g., the classic
6.1 Multiprocessor Computing Environment
6 Parallel Computing 6.1 Multiprocessor Computing Environment The high-performance computing environment used in this book for optimization of very large building structures is the Origin 2000 multiprocessor,
Module 14: "Directory-based Cache Coherence" Lecture 31: "Managing Directory Overhead" Directory-based Cache Coherence: Replacement of S blocks
Directory-based Cache Coherence: Replacement of S blocks Serialization VN deadlock Starvation Overflow schemes Sparse directory Remote access cache COMA Latency tolerance Page migration Queue lock in hardware
Computer Systems Architecture
Computer Systems Architecture Lecture 23 Mahadevan Gomathisankaran April 27, 2010 04/27/2010 Lecture 23 CSCE 4610/5610 1 Reminder ABET Feedback: http://www.cse.unt.edu/exitsurvey.cgi?csce+4610+001 Student
Chapter 9 Multiprocessors
ECE200 Computer Organization Chapter 9 Multiprocessors David H. lbonesi and the University of Rochester Henk Corporaal, TU Eindhoven, Netherlands Jari Nurmi, Tampere University of Technology, Finland University
Computer Systems Architecture
Computer Systems Architecture Lecture 24 Mahadevan Gomathisankaran April 29, 2010 04/29/2010 Lecture 24 CSCE 4610/5610 1 Reminder ABET Feedback: http://www.cse.unt.edu/exitsurvey.cgi?csce+4610+001 Student
Claude TADONKI. MINES ParisTech PSL Research University Centre de Recherche Informatique
Got 2 seconds Sequential 84 seconds Expected 84/84 = 1 second!?! Got 25 seconds MINES ParisTech PSL Research University Centre de Recherche Informatique claude.tadonki@mines-paristech.fr Séminaire MATHEMATIQUES
CSCI 402: Computer Architectures. Parallel Processors (2) Fengguang Song Department of Computer & Information Science IUPUI.
CSCI 402: Computer Architectures Parallel Processors (2) Fengguang Song Department of Computer & Information Science IUPUI 6.6 - End Today s Contents GPU Cluster and its network topology The Roofline performance
Experience the GRID Today with Oracle9i RAC
1 Experience the GRID Today with Oracle9i RAC Shig Hiura Pre-Sales Engineer Shig_Hiura@etagon.com 2 Agenda Introduction What is the Grid The Database Grid Oracle9i RAC Technology 10g vs. 9iR2 Comparison
HP OpenVMS Alpha and Integrity Performance Comparison
HP OpenVMS Alpha and Integrity Performance Comparison Session 1225 Gregory Jordan OpenVMS Engineering Hewlett-Packard 25 Hewlett-Packard Development Company, L.P. The information contained herein is subject
Distributed File Systems Issues. NFS (Network File System) AFS: Namespace. The Andrew File System (AFS) Operating Systems 11/19/2012 CSC 256/456 1
Distributed File Systems Issues NFS (Network File System) Naming and transparency (location transparency versus location independence) Host:local-name Attach remote directories (mount) Single global name
Performance Impact of Resource Contention in Multicore Systems
Performance Impact of Resource Contention in Multicore Systems R. Hood, H. Jin, P. Mehrotra, J. Chang, J. Djomehri, S. Gavali, D. Jespersen, K. Taylor, R. Biswas Commodity Multicore Chips in NASA HEC 2004:
Scalable Shared-Memory Implementations. Erik Hagersten Uppsala University
Scalable Shared-emory Implementations Erik Hagersten Uppsala University Sweden -to-cache in snoop-based A: B: BusRTS y RTS wait for data Gotta answer $ $ $ Thread Thread Thread Read A... Read A Read A
Issues in Multiprocessors
Issues in Multiprocessors Which programming model for interprocessor communication shared memory regular loads & stores message passing explicit sends & receives Which execution model control parallel
Lecture Topics. Announcements. Today: Advanced Scheduling (Stallings, chapter ) Next: Deadlock (Stallings, chapter
Lecture Topics Today: Advanced Scheduling (Stallings, chapter 10.1-10.4) Next: Deadlock (Stallings, chapter 6.1-6.6) 1 Announcements Exam #2 returned today Self-Study Exercise #10 Project #8 (due 11/16)
Scalable Cache Coherence
arallel Computing Scalable Cache Coherence Hwansoo Han Hierarchical Cache Coherence Hierarchies in cache organization Multiple levels of caches on a processor Large scale multiprocessors with hierarchy
Adapted from: TRENDS AND ATTRIBUTES OF HORIZONTAL AND VERTICAL COMPUTING ARCHITECTURES
Adapted from: TRENDS AND ATTRIBUTES OF HORIZONTAL AND VERTICAL COMPUTING ARCHITECTURES Tom Atwood Business Development Manager Sun Microsystems, Inc. Takeaways Understand the technical differences between
Handout 3 Multiprocessor and thread level parallelism
Handout 3 Multiprocessor and thread level parallelism Outline Review MP Motivation SISD v SIMD (SIMT) v MIMD Centralized vs Distributed Memory MESI and Directory Cache Coherency Synchronization and Relaxed
Introduction to parallel Computing
Introduction to parallel Computing VI-SEEM Training Paschalis Paschalis Korosoglou Korosoglou (pkoro@.gr) (pkoro@.gr) Outline Serial vs Parallel programming Hardware trends Why HPC matters HPC Concepts
Aleksandar Milenkovich 1
Parallel Computers Lecture 8: Multiprocessors Aleksandar Milenkovic, milenka@ece.uah.edu Electrical and Computer Engineering University of Alabama in Huntsville Definition: A parallel computer is a collection
Dheeraj Bhardwaj May 12, 2003
HPC Systems and Models Dheeraj Bhardwaj Department of Computer Science & Engineering Indian Institute of Technology, Delhi 110 016 India http://www.cse.iitd.ac.in/~dheerajb 1 Sequential Computers Traditional
Non-uniform memory access (NUMA)
Non-uniform memory access (NUMA) Memory access between processor core to main memory is not uniform. Memory resides in separate regions called NUMA domains. For highest performance, cores should only access
Review: Creating a Parallel Program. Programming for Performance
Review: Creating a Parallel Program Can be done by programmer, compiler, run-time system or OS Steps for creating parallel program Decomposition Assignment of tasks to processes Orchestration Mapping (C)
CSE502: Computer Architecture CSE 502: Computer Architecture
CSE 502: Computer Architecture Shared-Memory Multi-Processors Shared-Memory Multiprocessors Multiple threads use shared memory (address space) SysV Shared Memory or Threads in software Communication implicit
Performance study example ( 5.3) Performance study example
erformance study example ( 5.3) Coherence misses: - True sharing misses - Write to a shared block - ead an invalid block - False sharing misses - ead an unmodified word in an invalidated block CI for commercial
Exercises: April 11. Hermann Härtig, TU Dresden, Distributed OS, Load Balancing
Exercises: April 11 1 PARTITIONING IN MPI COMMUNICATION AND NOISE AS HPC BOTTLENECK LOAD BALANCING DISTRIBUTED OPERATING SYSTEMS, SCALABILITY, SS 2017 Hermann Härtig THIS LECTURE Partitioning: bulk synchronous
Parallel Computer Architecture Spring Distributed Shared Memory Architectures & Directory-Based Memory Coherence
Parallel Computer Architecture Spring 2018 Distributed Shared Memory Architectures & Directory-Based Memory Coherence Nikos Bellas Computer and Communications Engineering Department University of Thessaly
Non-uniform memory access machine or (NUMA) is a system where the memory access time to any region of memory is not the same for all processors.
CS 320 Ch. 17 Parallel Processing Multiple Processor Organization The author makes the statement: "Processors execute programs by executing machine instructions in a sequence one at a time." He also says
SMD149 - Operating Systems - Multiprocessing
SMD149 - Operating Systems - Multiprocessing Roland Parviainen December 1, 2005 1 / 55 Overview Introduction Multiprocessor systems Multiprocessor, operating system and memory organizations 2 / 55 Introduction
Overview. SMD149 - Operating Systems - Multiprocessing. Multiprocessing architecture. Introduction SISD. Flynn s taxonomy
Overview SMD149 - Operating Systems - Multiprocessing Roland Parviainen Multiprocessor systems Multiprocessor, operating system and memory organizations December 1, 2005 1/55 2/55 Multiprocessor system
Multiple Processor Systems. Lecture 15 Multiple Processor Systems. Multiprocessor Hardware (1) Multiprocessors. Multiprocessor Hardware (2)
Lecture 15 Multiple Processor Systems Multiple Processor Systems Multiprocessors Multicomputers Continuous need for faster computers shared memory model message passing multiprocessor wide area distributed
Commodity Cluster Computing
Commodity Cluster Computing Ralf Gruber, EPFL-SIC/CAPA/Swiss-Tx, Lausanne http://capawww.epfl.ch Commodity Cluster Computing 1. Introduction 2. Characterisation of nodes, parallel machines,applications
Agenda. System Performance Scaling of IBM POWER6 TM Based Servers
System Performance Scaling of IBM POWER6 TM Based Servers Jeff Stuecheli Hot Chips 19 August 2007 Agenda Historical background POWER6 TM chip components Interconnect topology Cache Coherence strategies
MD NASTRAN on Advanced SGI Architectures *
W h i t e P a p e r MD NASTRAN on Advanced SGI Architectures * Olivier Schreiber, Scott Shaw, Joe Griffin** Abstract MD Nastran tackles all important Normal Mode Analyses utilizing both Shared Memory Parallelism
MULTIPROCESSORS AND THREAD LEVEL PARALLELISM
UNIT III MULTIPROCESSORS AND THREAD LEVEL PARALLELISM 1. Symmetric Shared Memory Architectures: The Symmetric Shared Memory Architecture consists of several processors with a single physical memory shared
Chapter 5. Multiprocessors and Thread-Level Parallelism
Computer Architecture A Quantitative Approach, Fifth Edition Chapter 5 Multiprocessors and Thread-Level Parallelism 1 Introduction Thread-Level parallelism Have multiple program counters Uses MIMD model
Introduction to Parallel Computing
Portland State University ECE 588/688 Introduction to Parallel Computing Reference: Lawrence Livermore National Lab Tutorial https://computing.llnl.gov/tutorials/parallel_comp/ Copyright by Alaa Alameldeen
Page 1. SMP Review. Multiprocessors. Bus Based Coherence. Bus Based Coherence. Characteristics. Cache coherence. Cache coherence
SMP Review Multiprocessors Today s topics: SMP cache coherence general cache coherence issues snooping protocols Improved interaction lots of questions warning I m going to wait for answers granted it
Scaling to Petaflop. Ola Torudbakken Distinguished Engineer. Sun Microsystems, Inc
Scaling to Petaflop Ola Torudbakken Distinguished Engineer Sun Microsystems, Inc HPC Market growth is strong CAGR increased from 9.2% (2006) to 15.5% (2007) Market in 2007 doubled from 2003 (Source: IDC
NUMA-aware OpenMP Programming
NUMA-aware OpenMP Programming Dirk Schmidl IT Center, RWTH Aachen University Member of the HPC Group schmidl@itc.rwth-aachen.de Christian Terboven IT Center, RWTH Aachen University Deputy lead of the HPC
PARALLEL MEMORY ARCHITECTURE
PARALLEL MEMORY ARCHITECTURE Mahdi Nazm Bojnordi Assistant Professor School of Computing University of Utah CS/ECE 6810: Computer Architecture Overview Announcement Homework 6 is due tonight n The last
Software and Tools for HPE s The Machine Project
Labs Software and Tools for HPE s The Machine Project Scalable Tools Workshop Aug/1 - Aug/4, 2016 Lake Tahoe Milind Chabbi Traditional Computing Paradigm CPU DRAM CPU DRAM CPU-centric computing 2 CPU-Centric
Chapter 5. Multiprocessors and Thread-Level Parallelism
Computer Architecture A Quantitative Approach, Fifth Edition Chapter 5 Multiprocessors and Thread-Level Parallelism 1 Introduction Thread-Level parallelism Have multiple program counters Uses MIMD model
Computer parallelism Flynn s categories
04 Multi-processors 04.01-04.02 Taxonomy and communication Parallelism Taxonomy Communication alessandro bogliolo isti information science and technology institute 1/9 Computer parallelism Flynn s categories