INTERCONNECTION TECHNOLOGIES. Non-Uniform Memory Access Seminar Elina Zarisheva
|
|
|
- Sydney Higgins
- 5 years ago
- Views:
Transcription
1 INTERCONNECTION TECHNOLOGIES Non-Uniform Memory Access Seminar Elina Zarisheva
2 NUMA Seminar Elina Zarisheva 2 Agenda Network topology Logical vs. physical topology Logical topologies InfiniBand Crossbar switch Interconnection technologies in NUMA system AMD Hyper-Transport (HT) Intel Quick-Path Interconnect (QPI) NumaLink
3 NUMA Seminar Elina Zarisheva 3 Network Topology Computer system or network equipment are connected to each other Physical topologies Logical topologies
4 NUMA Seminar Elina Zarisheva 4 Logical topologies Point-to-point Bus Daisy chain Ring Star Mesh Tree Hybrid Hypercube
5 NUMA Seminar Elina Zarisheva 5 Point-to-point topology Connects two nodes directly together Types: simplex half-duplex full-duplex + simple + fast + medium is not shared - support only two nodes
6 NUMA Seminar Elina Zarisheva 6 Bus topology Common medium (central bus) where the rest of nodes separately connected + more than one node + costs - small networks - limitation of nodes - data collision - depended on central bus - security
7 NUMA Seminar Elina Zarisheva 7 Daisy chain topology Node connects one after another + simple + scalability - slow for the opposite end of the chain
8 NUMA Seminar Elina Zarisheva 8 Ring topology Each node has two connection: to its nearest neighbors Data transmission happens indirectly Sending and receiving data with the help of the token Clockwise Double ring + organized + no date collision + no server + easy to add components - slow - dependent T
9 NUMA Seminar Elina Zarisheva 9 Star topology Each node connects to a central point via a point-to-point link Central device: hub switch Independent access + centralized management + failure of a node does not affect - central devise failure affect all the network - cost - number of nodes depends on capacity of central device
10 NUMA Seminar Elina Zarisheva 10 Mesh topology Nodes are connected directly to another nodes Types: fully-connected partly-connected + simultaneously +failure of one node does not affect on the system + easy to modify - high redundancy - cost - set-up and administration is difficult
11 NUMA Seminar Elina Zarisheva 11 Tree (hierarchy) topology Star topology are connected using Bus topology + expansion is easy + easy to manage and maintain + error detection and correction are easy + failure of one segment does not affect to the system - central bus - maintenance becomes difficult - scalability depends on cable
12 NUMA Seminar Elina Zarisheva 12 Hybrid topology Combines two or more topologies Reap their advantages + scalable + flexible + effective - complex - cost
13 NUMA Seminar Elina Zarisheva 13 Hypercube topology The distance between any two nodes is at most log(p) Each node has log(p) neighbors + only log(p) neighbors + the longest route is log(p) + excellent connectivity - physical network infrastructure is ignored - building one node at a time
14 NUMA Seminar Elina Zarisheva 14 InfiniBand (IBA) High bandwidth Low latency Bi-directional Shared-nothing architecture Not able to address directly Support for up to 64,000 addressable devices
15 NUMA Seminar Elina Zarisheva 15 IBA bandwidth 1X Link 500MB/sec 4X Link 2GB/sec 12X Link 6 GB/sec
16 IBA System Fabric NUMA Seminar Elina Zarisheva 16
17 NUMA Seminar Elina Zarisheva 17 IBA Used by 43% of the systems on the Top-500 list of supercomputers (Nov. 2010) Intel Xeon Phi Mellanox ConnectX IB
18 NUMA Seminar Elina Zarisheva 18 Crossbar switch (Integrity Superdome 2)
19 INTERCONNECTION TECHNOLOGIES IN NUMA
20 NUMA Seminar Elina Zarisheva 20 Non-Uniform Memory Access CPU CPU CPU CPU Node Interconnect Node Memory controller Node Node
21 NUMA Seminar Elina Zarisheva 21 Interconnectors Node Interconnect Node CPU CPU CPU CPU Memory controller Node Node
22 NUMA Seminar Elina Zarisheva 22 NUMA Interconnectors AMD Hyper-Transport (HT) Intel Quick-Path Interconnect (QPI) NumaLink
23 NUMA Seminar Elina Zarisheva 23 AMD Hyper-Transport (HT) High bandwidth Low latency Bi-directionally Multiple configuration Daisy chain Star Mesh HyperTransport 3.10
24 NUMA Seminar Elina Zarisheva 24 AMD Hyper-Transport (HT) Minimal chain: two nodes Maximal chain: 32 nodes Up to 3.2 GB/sec over 8-bit I/O link Up to 12.8 GB/sec over 16-bit I/O link Up to 25.6 GB/sec over 32-bit I/O link
25 HT Implementation NUMA Seminar Elina Zarisheva 25
26 NUMA Seminar Elina Zarisheva 26 AMD Hyper-Transport (HT) Opteron Athlon 64 Phenom
27 NUMA Seminar Elina Zarisheva 27 Intel Quick-Path Interconnect (QPI) High bandwidth Low latency Bi-directional Up to GB/sec per link High scalable
28 NUMA Seminar Elina Zarisheva 28 QPI Physical Layer Two 20-lane point-to-point data links Full duplex with a separate clock pair in each direction 42 signals Total number of pins is data lanes are divided onto four "quadrants" of 5 lanes each
29 NUMA Seminar Elina Zarisheva 29 Intel Quick-Path Interconnect (QPI)
30 NUMA Seminar Elina Zarisheva 30 Intel Quick-Path Interconnect (QPI) Xeon Core i7 Itanium
31 NUMA Seminar Elina Zarisheva 31 SGI Numalink High bandwidth Low latency Bi-directional Connected into the memory infrastructure of the system Reduce one-way request to memory Numalink3 3.2 GB/sec Numalink4 6.4 GB/sec Different topology Max 2048 nodes
32 NUMA Seminar Elina Zarisheva 32 SGI Numalink. Tree configuration
33 NUMA Seminar Elina Zarisheva 33 SGI UV Hypercube configuration
34 NUMA Seminar Elina Zarisheva 34 SGI Numalink. Hypercube configuration
35 NUMA Seminar Elina Zarisheva 35 SGI Numalink Altix servers and supercomputers
36 NUMA Seminar Elina Zarisheva 36 Questions?
Parallel Architectures
Parallel Architectures Part 1: The rise of parallel machines Intel Core i7 4 CPU cores 2 hardware thread per core (8 cores ) Lab Cluster Intel Xeon 4/10/16/18 CPU cores 2 hardware thread per core (8/20/32/36
COSC 6374 Parallel Computation. Parallel Computer Architectures
OS 6374 Parallel omputation Parallel omputer Architectures Some slides on network topologies based on a similar presentation by Michael Resch, University of Stuttgart Spring 2010 Flynn s Taxonomy SISD:
COSC 6374 Parallel Computation. Parallel Computer Architectures
OS 6374 Parallel omputation Parallel omputer Architectures Some slides on network topologies based on a similar presentation by Michael Resch, University of Stuttgart Edgar Gabriel Fall 2015 Flynn s Taxonomy
CS 770G - Parallel Algorithms in Scientific Computing Parallel Architectures. May 7, 2001 Lecture 2
CS 770G - arallel Algorithms in Scientific Computing arallel Architectures May 7, 2001 Lecture 2 References arallel Computer Architecture: A Hardware / Software Approach Culler, Singh, Gupta, Morgan Kaufmann
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
Designing High Performance Communication Middleware with Emerging Multi-core Architectures
Designing High Performance Communication Middleware with Emerging Multi-core Architectures Dhabaleswar K. (DK) Panda Department of Computer Science and Engg. The Ohio State University E-mail: panda@cse.ohio-state.edu
Interconnection Network. Jinkyu Jeong Computer Systems Laboratory Sungkyunkwan University
Interconnection Network Jinkyu Jeong (jinkyu@skku.edu) Computer Systems Laboratory Sungkyunkwan University http://csl.skku.edu Topics Taxonomy Metric Topologies Characteristics Cost Performance 2 Interconnection
Interconnection Network
Interconnection Network Jinkyu Jeong (jinkyu@skku.edu) Computer Systems Laboratory Sungkyunkwan University http://csl.skku.edu SSE3054: Multicore Systems, Spring 2017, Jinkyu Jeong (jinkyu@skku.edu) Topics
CS Parallel Algorithms in Scientific Computing
CS 775 - arallel Algorithms in Scientific Computing arallel Architectures January 2, 2004 Lecture 2 References arallel Computer Architecture: A Hardware / Software Approach Culler, Singh, Gupta, Morgan
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]
Different network topologies
Network Topology Network topology is the arrangement of the various elements of a communication network. It is the topological structure of a network and may be depicted physically or logically. Physical
Network Infrastructure
Network Infrastructure For building computer networks more complex than e.g. a short bus, some additional components are needed. They can be arranged hierarchically regarding their functionality: Repeater
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
Designing Multi-Leader-Based Allgather Algorithms for Multi-Core Clusters *
Designing Multi-Leader-Based Allgather Algorithms for Multi-Core Clusters * Krishna Kandalla, Hari Subramoni, Gopal Santhanaraman, Matthew Koop and Dhabaleswar K. Panda Department of Computer Science and
UNIT 5 MANAGING COMPUTER NETWORKS LEVEL 3 NETWORK TOPOLOGIES AND LAYOUT
UNIT 5 MANAGING COMPUTER NETWORKS LEVEL 3 NETWORK TOPOLOGIES AND LAYOUT NETWORK TOPOLOGY Network Topology refers to layout of a network and how different nodes in a network are connected to each other
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
HPC Architectures. Types of resource currently in use
HPC Architectures Types of resource currently in use Reusing this material This work is licensed under a Creative Commons Attribution- NonCommercial-ShareAlike 4.0 International License. http://creativecommons.org/licenses/by-nc-sa/4.0/deed.en_us
Interconnection Networks. Issues for Networks
Interconnection Networks Communications Among Processors Chris Nevison, Colgate University Issues for Networks Total Bandwidth amount of data which can be moved from somewhere to somewhere per unit time
Lecture 2 Parallel Programming Platforms
Lecture 2 Parallel Programming Platforms Flynn s Taxonomy In 1966, Michael Flynn classified systems according to numbers of instruction streams and the number of data stream. Data stream Single Multiple
Memory Scalability Evaluation of the Next-Generation Intel Bensley Platform with InfiniBand
Memory Scalability Evaluation of the Next-Generation Intel Bensley Platform with InfiniBand Matthew Koop, Wei Huang, Ahbinav Vishnu, Dhabaleswar K. Panda Network-Based Computing Laboratory Department of
Computing architectures Part 2 TMA4280 Introduction to Supercomputing
Computing architectures Part 2 TMA4280 Introduction to Supercomputing NTNU, IMF January 16. 2017 1 Supercomputing What is the motivation for Supercomputing? Solve complex problems fast and accurately:
Fundamentals of Networking Types of Topologies
Fundamentals of Networking Types of Topologies Kuldeep Sonar 1 Bus Topology Bus topology is a network type in which every computer and network device is connected to single cable. When it has exactly two
MELLANOX EDR UPDATE & GPUDIRECT MELLANOX SR. SE 정연구
MELLANOX EDR UPDATE & GPUDIRECT MELLANOX SR. SE 정연구 Leading Supplier of End-to-End Interconnect Solutions Analyze Enabling the Use of Data Store ICs Comprehensive End-to-End InfiniBand and Ethernet Portfolio
Parallel Computing Platforms
Parallel Computing Platforms Network Topologies John Mellor-Crummey Department of Computer Science Rice University johnmc@rice.edu COMP 422/534 Lecture 14 28 February 2017 Topics for Today Taxonomy Metrics
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
Some previous important concepts
Almustansorya University College of Education Computer Science Department Communication and Computer Networks Class 4 (A/B) Lesson 6... Network Topology Abstract This lesson will mainly focus on the network
Application Performance on Dual Processor Cluster Nodes
Application Performance on Dual Processor Cluster Nodes by Kent Milfeld milfeld@tacc.utexas.edu edu Avijit Purkayastha, Kent Milfeld, Chona Guiang, Jay Boisseau TEXAS ADVANCED COMPUTING CENTER Thanks Newisys
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)
Interconnect Technology and Computational Speed
Interconnect Technology and Computational Speed From Chapter 1 of B. Wilkinson et al., PARAL- LEL PROGRAMMING. Techniques and Applications Using Networked Workstations and Parallel Computers, augmented
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
Improving Linear Algebra Computation on NUMA platforms through auto-tuned tuned nested parallelism
Improving Linear Algebra Computation on NUMA platforms through auto-tuned tuned nested parallelism Javier Cuenca, Luis P. García, Domingo Giménez Parallel Computing Group University of Murcia, SPAIN parallelum
Physical Organization of Parallel Platforms. Alexandre David
Physical Organization of Parallel Platforms Alexandre David 1.2.05 1 Static vs. Dynamic Networks 13-02-2008 Alexandre David, MVP'08 2 Interconnection networks built using links and switches. How to connect:
CS575 Parallel Processing
CS575 Parallel Processing Lecture three: Interconnection Networks Wim Bohm, CSU Except as otherwise noted, the content of this presentation is licensed under the Creative Commons Attribution 2.5 license.
CCS HPC. Interconnection Network. PC MPP (Massively Parallel Processor) MPP IBM
CCS HC taisuke@cs.tsukuba.ac.jp 1 2 CU memoryi/o 2 2 4single chipmulti-core CU 10 C CM (Massively arallel rocessor) M IBM BlueGene/L 65536 Interconnection Network 3 4 (distributed memory system) (shared
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:
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
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
Outline. Distributed Shared Memory. Shared Memory. ECE574 Cluster Computing. Dichotomy of Parallel Computing Platforms (Continued)
Cluster Computing Dichotomy of Parallel Computing Platforms (Continued) Lecturer: Dr Yifeng Zhu Class Review Interconnections Crossbar» Example: myrinet Multistage» Example: Omega network Outline Flynn
SGI UV for SAP HANA. Scale-up, Single-node Architecture Enables Real-time Operations at Extreme Scale and Lower TCO
W h i t e P a p e r SGI UV for SAP HANA Scale-up, Single-node Architecture Enables Real-time Operations at Extreme Scale and Lower TCO Table of Contents Introduction 1 SGI UV for SAP HANA 1 Architectural
Virtual Leverage: Server Consolidation in Open Source Environments. Margaret Lewis Commercial Software Strategist AMD
Virtual Leverage: Server Consolidation in Open Source Environments Margaret Lewis Commercial Software Strategist AMD What Is Virtualization? Abstraction of Hardware Components Virtual Memory Virtual Volume
Area Covered is small Area covered is large. Data transfer rate is high Data transfer rate is low
Chapter 15 Networking Concepts 1. Define networking. It is the interconnection of independent computing devices for sharing of information over shared medium. 2. What is the need for networking? / What
Performance Optimizations via Connect-IB and Dynamically Connected Transport Service for Maximum Performance on LS-DYNA
Performance Optimizations via Connect-IB and Dynamically Connected Transport Service for Maximum Performance on LS-DYNA Pak Lui, Gilad Shainer, Brian Klaff Mellanox Technologies Abstract From concept to
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,
Chapter 6 Connecting Device
Computer Networks Al-Mustansiryah University Elec. Eng. Department College of Engineering Fourth Year Class Chapter 6 Connecting Device 6.1 Functions of network devices Separating (connecting) networks
Interconnect Performance Evaluation of SGI Altix Cray X1, Cray Opteron, and Dell PowerEdge
Interconnect Performance Evaluation of SGI Altix 3700 Cray X1, Cray Opteron, and Dell PowerEdge Panagiotis Adamidis University of Stuttgart High Performance Computing Center, Nobelstrasse 19, D-70569 Stuttgart,
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
4. Networks. in parallel computers. Advances in Computer Architecture
4. Networks in parallel computers Advances in Computer Architecture System architectures for parallel computers Control organization Single Instruction stream Multiple Data stream (SIMD) All processors
What is Parallel Computing?
What is Parallel Computing? Parallel Computing is several processing elements working simultaneously to solve a problem faster. 1/33 What is Parallel Computing? Parallel Computing is several processing
Lecture 19. Optimizing for the memory hierarchy NUMA
Lecture 19 Optimizing for the memory hierarchy NUMA A Performance Puzzle When we fuse the ODE and PDE loops in A3, we slow down the code But, we actually observe fewer memory reads and cache misses! What
EN2910A: Advanced Computer Architecture Topic 06: Supercomputers & Data Centers Prof. Sherief Reda School of Engineering Brown University
EN2910A: Advanced Computer Architecture Topic 06: Supercomputers & Data Centers Prof. Sherief Reda School of Engineering Brown University Material from: The Datacenter as a Computer: An Introduction to
CSC630/CSC730: Parallel Computing
CSC630/CSC730: Parallel Computing Parallel Computing Platforms Chapter 2 (2.4.1 2.4.4) Dr. Joe Zhang PDC-4: Topology 1 Content Parallel computing platforms Logical organization (a programmer s view) Control
Data Communication. Introduction of Communication. Data Communication. Elements of Data Communication (Communication Model)
Data Communication Introduction of Communication The need to communicate is part of man s inherent being. Since the beginning of time the human race has communicated using different techniques and methods.
Parallel Processors. The dream of computer architects since 1950s: replicate processors to add performance vs. design a faster processor
Multiprocessing Parallel Computers Definition: A parallel computer is a collection of processing elements that cooperate and communicate to solve large problems fast. Almasi and Gottlieb, Highly Parallel
Interconnect Your Future
Interconnect Your Future Gilad Shainer 2nd Annual MVAPICH User Group (MUG) Meeting, August 2014 Complete High-Performance Scalable Interconnect Infrastructure Comprehensive End-to-End Software Accelerators
Module 2 Storage Network Architecture
Module 2 Storage Network Architecture 1. SCSI 2. FC Protocol Stack 3. SAN:FC SAN 4. IP Storage 5. Infiniband and Virtual Interfaces FIBRE CHANNEL SAN 1. First consider the three FC topologies pointto-point,
Chapter 1. Introduction: Part I. Jens Saak Scientific Computing II 7/348
Chapter 1 Introduction: Part I Jens Saak Scientific Computing II 7/348 Why Parallel Computing? 1. Problem size exceeds desktop capabilities. Jens Saak Scientific Computing II 8/348 Why Parallel Computing?
CS395T: Introduction to Scientific and Technical Computing
CS395T: Introduction to Scientific and Technical Computing Instructors: Dr. Karl W. Schulz, Research Associate, TACC Dr. Bill Barth, Research Associate, TACC 1 Outline Parallel Computer Architectures Components
ABySS Performance Benchmark and Profiling. May 2010
ABySS Performance Benchmark and Profiling May 2010 Note The following research was performed under the HPC Advisory Council activities Participating vendors: AMD, Dell, Mellanox Compute resource - HPC
COMPUTER NETWORKING. By: Dr. Noor Dayana Abd Halim
COMPUTER NETWORKING By: Dr. Noor Dayana Abd Halim Defining Computer Network Computer network is a collection of computers and other hardware devices so that network users can share hardware, software,
Advanced Parallel Architecture. Annalisa Massini /2017
Advanced Parallel Architecture Annalisa Massini - 2016/2017 References Advanced Computer Architecture and Parallel Processing H. El-Rewini, M. Abd-El-Barr, John Wiley and Sons, 2005 Parallel computing
Types of Computer Networks and their Topologies Three important groups of computer networks: LAN, MAN, WAN
Types of Computer and their Topologies Three important groups of computer networks: LAN, MAN, WAN LAN (Local Area ) 1 MAN (Metropolitan Area ) 2 WAN (Wide Area ) 3 Problems to be discussed when presenting
ISO/OSI Model and Collision Domain NETWORK INFRASTRUCTURES NETKIT - LECTURE 1 MANUEL CAMPO, MARCO SPAZIANI
ISO/OSI Model and Collision Domain NETWORK INFRASTRUCTURES NETKIT - LECTURE 1 MANUEL CAMPO, MARCO SPAZIANI ISO/OSI Model ISO: International Organization for Standardization OSI: Open Systems Interconnection
SGI UV 300RL for Oracle Database In-Memory
SGI UV 300RL for Oracle Database In- Single-system Architecture Enables Real-time Business at Near Limitless Scale with Mission-critical Reliability TABLE OF CONTENTS 1.0 Introduction 1 2.0 SGI In- Computing
SGI Overview. HPC User Forum Dearborn, Michigan September 17 th, 2012
SGI Overview HPC User Forum Dearborn, Michigan September 17 th, 2012 SGI Market Strategy HPC Commercial Scientific Modeling & Simulation Big Data Hadoop In-memory Analytics Archive Cloud Public Private
HyperTransport. Dennis Vega Ryan Rawlins
HyperTransport Dennis Vega Ryan Rawlins What is HyperTransport (HT)? A point to point interconnect technology that links processors to other processors, coprocessors, I/O controllers, and peripheral controllers.
Intel QuickPath Interconnect Electrical Architecture Overview
Chapter 1 Intel QuickPath Interconnect Electrical Architecture Overview The art of progress is to preserve order amid change and to preserve change amid order Alfred North Whitehead The goal of this chapter
Chapter 4 NETWORK HARDWARE
Chapter 4 NETWORK HARDWARE 1 Network Devices As Organizations grow, so do their networks Growth in number of users Geographical Growth Network Devices : Are products used to expand or connect networks.
Lecture: Interconnection Networks
Lecture: Interconnection Networks Topics: Router microarchitecture, topologies Final exam next Tuesday: same rules as the first midterm 1 Packets/Flits A message is broken into multiple packets (each packet
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
HPC Solution. Technology for a New Era in Computing
HPC Solution Technology for a New Era in Computing TEL IN HPC & Storage.. 20 years of changing with Technology Complete Solution Integrators for Select Verticals Mechanical Design & Engineering High Performance
Lecture 26: Interconnects. James C. Hoe Department of ECE Carnegie Mellon University
18 447 Lecture 26: Interconnects James C. Hoe Department of ECE Carnegie Mellon University 18 447 S18 L26 S1, James C. Hoe, CMU/ECE/CALCM, 2018 Housekeeping Your goal today get an overview of parallel
Lecture 12: Interconnection Networks. Topics: communication latency, centralized and decentralized switches, routing, deadlocks (Appendix E)
Lecture 12: Interconnection Networks Topics: communication latency, centralized and decentralized switches, routing, deadlocks (Appendix E) 1 Topologies Internet topologies are not very regular they grew
COMP4300/8300: Overview of Parallel Hardware. Alistair Rendell. COMP4300/8300 Lecture 2-1 Copyright c 2015 The Australian National University
COMP4300/8300: Overview of Parallel Hardware Alistair Rendell COMP4300/8300 Lecture 2-1 Copyright c 2015 The Australian National University 2.1 Lecture Outline Review of Single Processor Design So we talk
The AMD64 Technology for Server and Workstation. Dr. Ulrich Knechtel Enterprise Program Manager EMEA
The AMD64 Technology for Server and Workstation Dr. Ulrich Knechtel Enterprise Program Manager EMEA Agenda Direct Connect Architecture AMD Opteron TM Processor Roadmap Competition OEM support The AMD64
COMP 322: Principles of Parallel Programming. Lecture 17: Understanding Parallel Computers (Chapter 2) Fall 2009
COMP 322: Principles of Parallel Programming Lecture 17: Understanding Parallel Computers (Chapter 2) Fall 2009 http://www.cs.rice.edu/~vsarkar/comp322 Vivek Sarkar Department of Computer Science Rice
Networks. Distributed Systems. Philipp Kupferschmied. Universität Karlsruhe, System Architecture Group. May 6th, 2009
Networks Distributed Systems Philipp Kupferschmied Universität Karlsruhe, System Architecture Group May 6th, 2009 Philipp Kupferschmied Networks 1/ 41 1 Communication Basics Introduction Layered Communication
S8688 : INSIDE DGX-2. Glenn Dearth, Vyas Venkataraman Mar 28, 2018
S8688 : INSIDE DGX-2 Glenn Dearth, Vyas Venkataraman Mar 28, 2018 Why was DGX-2 created Agenda DGX-2 internal architecture Software programming model Simple application Results 2 DEEP LEARNING TRENDS Application
LS-DYNA Productivity and Power-aware Simulations in Cluster Environments
LS-DYNA Productivity and Power-aware Simulations in Cluster Environments Gilad Shainer 1, Tong Liu 1, Jacob Liberman 2, Jeff Layton 2 Onur Celebioglu 2, Scot A. Schultz 3, Joshua Mora 3, David Cownie 3,
AMD Opteron Processor. Architectures for Multimedia Systems A.Y. 2009/2010 Simone Segalini
AMD Opteron Processor Architectures for Multimedia Systems A.Y. 2009/2010 Simone Segalini A brief of history Released on April 22, 2003 (codename SledgeHammer) First processor to implement AMD64 instruction
Overview. Processor organizations Types of parallel machines. Real machines
Course Outline Introduction in algorithms and applications Parallel machines and architectures Overview of parallel machines, trends in top-500, clusters, DAS Programming methods, languages, and environments
Modern computer architecture. From multicore to petaflops
Modern computer architecture From multicore to petaflops Motivation: Multi-ores where and why Introduction: Moore s law Intel Sandy Brige EP: 2.3 Billion nvidia FERMI: 3 Billion 1965: G. Moore claimed
2008 International ANSYS Conference
28 International ANSYS Conference Maximizing Performance for Large Scale Analysis on Multi-core Processor Systems Don Mize Technical Consultant Hewlett Packard 28 ANSYS, Inc. All rights reserved. 1 ANSYS,
Parallel Computing Platforms
Parallel Computing Platforms Routing, Network Embedding John Mellor-Crummey Department of Computer Science Rice University johnmc@rice.edu COMP 422/534 Lecture 14-15 4,11 October 2018 Topics for Today
Project: IEEE P Working Group for Wireless Personal Area Networks N
Project: IEEE P802.15 Working Group for Wireless Personal Area Networks N (WPANs( WPANs) Submission Title: [Technical Requirements for VLC Applications] Date Submitted: [March, 2009] Source: [Dae-Ho Kim,
Future Routing Schemes in Petascale clusters
Future Routing Schemes in Petascale clusters Gilad Shainer, Mellanox, USA Ola Torudbakken, Sun Microsystems, Norway Richard Graham, Oak Ridge National Laboratory, USA Birds of a Feather Presentation Abstract
Communication has significant impact on application performance. Interconnection networks therefore have a vital role in cluster systems.
Cluster Networks Introduction Communication has significant impact on application performance. Interconnection networks therefore have a vital role in cluster systems. As usual, the driver is performance
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
Computer Networks รศ.ดร.อน นต ผลเพ ม. Assoc. Prof. Anan Phonphoem, Ph.D. Kasetsart University, Bangkok, Thailand
Jan May 2018 Computer Networks รศ.ดร.อน นต ผลเพ ม Assoc. Prof. Anan Phonphoem, Ph.D. anan.p@ku.ac.th http://www.cpe.ku.ac.th/~anan Computer Engineering Department Kasetsart University, Bangkok, Thailand
Low latency, high bandwidth communication. Infiniband and RDMA programming. Bandwidth vs latency. Knut Omang Ifi/Oracle 2 Nov, 2015
Low latency, high bandwidth communication. Infiniband and RDMA programming Knut Omang Ifi/Oracle 2 Nov, 2015 1 Bandwidth vs latency There is an old network saying: Bandwidth problems can be cured with
Chapter 16 Networking
Chapter 16 Networking Outline 16.1 Introduction 16.2 Network Topology 16.3 Network Types 16.4 TCP/IP Protocol Stack 16.5 Application Layer 16.5.1 Hypertext Transfer Protocol (HTTP) 16.5.2 File Transfer
Single-Points of Performance
Single-Points of Performance Mellanox Technologies Inc. 29 Stender Way, Santa Clara, CA 9554 Tel: 48-97-34 Fax: 48-97-343 http://www.mellanox.com High-performance computations are rapidly becoming a critical
TDT Appendix E Interconnection Networks
TDT 4260 Appendix E Interconnection Networks Review Advantages of a snooping coherency protocol? Disadvantages of a snooping coherency protocol? Advantages of a directory coherency protocol? Disadvantages
HYCOM Performance Benchmark and Profiling
HYCOM Performance Benchmark and Profiling Jan 2011 Acknowledgment: - The DoD High Performance Computing Modernization Program Note The following research was performed under the HPC Advisory Council activities
NETWORK TOPOLOGIES. Application Notes. Keywords Topology, P2P, Bus, Ring, Star, Mesh, Tree, PON, Ethernet. Author John Peter & Timo Perttunen
Application Notes NETWORK TOPOLOGIES Author John Peter & Timo Perttunen Issued June 2014 Abstract Network topology is the way various components of a network (like nodes, links, peripherals, etc) are arranged.
COMP Parallel Computing. SMM (1) Memory Hierarchies and Shared Memory
COMP 633 - Parallel Computing Lecture 6 September 6, 2018 SMM (1) Memory Hierarchies and Shared Memory 1 Topics Memory systems organization caches and the memory hierarchy influence of the memory hierarchy
Introduction Electrical Considerations Data Transfer Synchronization Bus Arbitration VME Bus Local Buses PCI Bus PCI Bus Variants Serial Buses
Introduction Electrical Considerations Data Transfer Synchronization Bus Arbitration VME Bus Local Buses PCI Bus PCI Bus Variants Serial Buses 1 Most of the integrated I/O subsystems are connected to the
Christian Brothers High School, Lewisham. Year 12 Information Processes & Technology. Assessment Task 3: Communications Systems.
Name: Teacher: Christian Brothers High School, Lewisham Year 12 Information Processes & Technology Assessment Task 3: Communications Systems June 2001 Outcomes Assessed: H1.1, H1.2, H2.1, H3.1, H5.2. Weighting:
Emerging Protocols & Applications
Emerging Protocols & Applications Anthony Dalleggio Executive Vice President Modelware, Inc. Contents Historical System Bandwidth Trends The PC Bus SONET, UTOPIA, & POS-PHY Targeted Applications SANs:
cs/ee 143 Fall
cs/ee 143 Fall 2018 5 2 Ethernet 2.1 W&P, P3.2 3 Points. Consider the Slotted ALOHA MAC protocol. There are N nodes sharing a medium, and time is divided into slots. Each packet takes up a single slot.
Parallel Systems Prof. James L. Frankel Harvard University. Version of 6:50 PM 4-Dec-2018 Copyright 2018, 2017 James L. Frankel. All rights reserved.
Parallel Systems Prof. James L. Frankel Harvard University Version of 6:50 PM 4-Dec-2018 Copyright 2018, 2017 James L. Frankel. All rights reserved. Architectures SISD (Single Instruction, Single Data)