1 Interconnection Networks Direct Indirect Shared Memory Distributed Memory (Message passing)

Slides:



Advertisements
Similar presentations
Comparison Of Network On Chip Topologies Ahmet Salih BÜYÜKKAYHAN Fall.
Advertisements

Shantanu Dutt Univ. of Illinois at Chicago
Parallel Architectures: Topologies Heiko Schröder, 2003.
Parallel Architectures: Topologies Heiko Schröder, 2003.
CSCI 8150 Advanced Computer Architecture Hwang, Chapter 2 Program and Network Properties 2.4 System Interconnect Architectures.
1 Lecture 23: Interconnection Networks Topics: communication latency, centralized and decentralized switches (Appendix E)
Advanced Topics in Algorithms and Data Structures An overview of the lecture 2 Models of parallel computation Characteristics of SIMD models Design issue.
ECE669 L12: Interconnection Network Performance March 9, 2004 ECE 669 Parallel Computer Architecture Lecture 12 Interconnection Network Performance.
Interconnection Networks 1 Interconnection Networks (Chapter 6) References: [1,Wilkenson and Allyn, Ch. 1] [2, Akl, Chapter 2] [3, Quinn, Chapter 2-3]
Parallel Routing Bruce, Chiu-Wing Sham. Overview Background Routing in parallel computers Routing in hypercube network –Bit-fixing routing algorithm –Randomized.
ICN’s The n-D hypercube (n-cube) contains 2^n nodes (processors).
Interconnection Network PRAM Model is too simple Physically, PEs communicate through the network (either buses or switching networks) Cost depends on network.
Interconnection Networks  Lecture 5 : January 29 th 2007 Prof. Chung-Kuan Cheng University of California San Diego Transcribed by: Jason Thurkettle.
1 Tuesday, September 26, 2006 Wisdom consists of knowing when to avoid perfection. -Horowitz.
ECE669 L25: Final Exam Review May 6, 2004 ECE 669 Parallel Computer Architecture Lecture 25 Final Exam Review.
Models of Parallel Computation Advanced Algorithms & Data Structures Lecture Theme 12 Prof. Dr. Th. Ottmann Summer Semester 2006.
Interconnection Network Topologies
Interconnection Network Topology Design Trade-offs
1 Lecture 25: Interconnection Networks Topics: communication latency, centralized and decentralized switches, routing, deadlocks (Appendix E) Review session,
CS252/Patterson Lec /28/01 CS162 Computer Architecture Lecture 16: Multiprocessor 2: Directory Protocol, Interconnection Networks.
1 Static Interconnection Networks CEG 4131 Computer Architecture III Miodrag Bolic.
ECE669 L16: Interconnection Topology March 30, 2004 ECE 669 Parallel Computer Architecture Lecture 16 Interconnection Topology.
John Kubiatowicz Electrical Engineering and Computer Sciences
Interconnect Network Topologies
CS252 Graduate Computer Architecture Lecture 15 Multiprocessor Networks March 14 th, 2011 John Kubiatowicz Electrical Engineering and Computer Sciences.
CS252 Graduate Computer Architecture Lecture 15 Multiprocessor Networks March 12 th, 2012 John Kubiatowicz Electrical Engineering and Computer Sciences.
Interconnection Networks. Applications of Interconnection Nets Interconnection networks are used everywhere! ◦ Supercomputers – connecting the processors.
Blue Gene / C Cellular architecture 64-bit Cyclops64 chip: –500 Mhz –80 processors ( each has 2 thread units and a FP unit) Software –Cyclops64 exposes.
Course Outline Introduction in software and applications. Parallel machines and architectures –Overview of parallel machines –Cluster computers (Myrinet)
Interconnect Networks
Network Topologies Topology – how nodes are connected – where there is a wire between 2 nodes. Routing – the path a message takes to get from one node.
CS668- Lecture 2 - Sept. 30 Today’s topics Parallel Architectures (Chapter 2) Memory Hierarchy Busses and Switched Networks Interconnection Network Topologies.
CSE Advanced Computer Architecture Week-11 April 1, 2004 engr.smu.edu/~rewini/8383.
Minimal Fault Diameter for Highly Resilient Product Networks Khaled Day, Abdel-Elah Al-Ayyoub IEEE Trans. On Parallel and Distributed Systems 2000 vol.
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display. Parallel Programming in C with MPI and OpenMP Michael J. Quinn.
Course Wrap-Up Miodrag Bolic CEG4136. What was covered Interconnection network topologies and performance Shared-memory architectures Message passing.
Lecture 3 Innerconnection Networks for Parallel Computers
Network-on-Chip Introduction Axel Jantsch / Ingo Sander
Anshul Kumar, CSE IITD CSL718 : Multiprocessors Interconnection Mechanisms Performance Models 20 th April, 2006.
Computer Science and Engineering Parallel and Distributed Processing CSE 8380 January Session 4.
1 Lecture 13: LRC & Interconnection Networks Topics: LRC implementation, interconnection characteristics.
InterConnection Network Topologies to Minimize graph diameter: Low Diameter Regular graphs and Physical Wire Length Constrained networks Nilesh Choudhury.
Embedding long paths in k-ary n-cubes with faulty nodes and links
Anshul Kumar, CSE IITD ECE729 : Advanced Computer Architecture Lecture 27, 28: Interconnection Mechanisms In Multiprocessors 29 th, 31 st March, 2010.
Birds Eye View of Interconnection Networks
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display. Shared versus Switched Media.
1 Interconnection Networks. 2 Interconnection Networks Interconnection Network (for SIMD/MIMD) can be used for internal connections among: Processors,
Computer Science and Engineering Copyright by Hesham El-Rewini Advanced Computer Architecture.
Super computers Parallel Processing
Computer Science and Engineering Copyright by Hesham El-Rewini Advanced Computer Architecture CSE 8383 February Session 9.
HYPERCUBE ALGORITHMS-1
Topology How the components are connected. Properties Diameter Nodal degree Bisection bandwidth A good topology: small diameter, small nodal degree, large.
Spring EE 437 Lillevik 437s06-l22 University of Portland School of Engineering Advanced Computer Architecture Lecture 22 Distributed computer Interconnection.
1 Lecture 14: Interconnection Networks Topics: dimension vs. arity, deadlock.
Parallel Processing & Distributed Systems Thoai Nam Chapter 3.
Interconnection Networks Communications Among Processors.
Distributed and Parallel Processing
Lecture 23: Interconnection Networks
Connection System Serve on mutual connection processors and memory .
Interconnection topologies
John Kubiatowicz Electrical Engineering and Computer Sciences
Butterfly Network A butterfly network consists of (K+1)2^k nodes divided into K+1 Rows, or Ranks. Let node (i,j) refer to the jth node in the ith Rank.
Interconnection Network Design Lecture 14
Static Interconnection Networks
High Performance Computing & Bioinformatics Part 2 Dr. Imad Mahgoub
Interconnection Networks Contd.
Embedded Computer Architecture 5SAI0 Interconnection Networks
Interconnection Networks
ECE 753: FAULT-TOLERANT COMPUTING
Static Interconnection Networks
Presentation transcript:

1 Interconnection Networks Direct Indirect Shared Memory Distributed Memory (Message passing)

2 Topology Diameter: Longest path length between two processors d = max {d min (v i, v j ) | for all v i, v j  V} Node connectivity min. # of nodes to be removed for the network to be disconnected node connectivity r  There are r disjoint paths between every pair of nodes.

3 Topology Fault diameter Suppose the node connectivity = r, diameter of the network with at most r-1 faulty nodes Bisection width The # of edges to be removed to separate the graph into two equal parts. Cost 1) # of nodes & # of edges 2) Bisection width.

4 cw (Channel width): # of wires connecting two nodes. cr (Channel rate): # of bits/sec/wire. Channel bandwidth = cw  cr Bisection bandwidth = bisection width  channel bandwidth Cost  Bisection bandwidth bisection bandwidth is constant

5 Hypercube: N = 2 2n bisection width = 2 2n-1 channel width = w h bisection bandwidth = 2 2n-1  w h 2-D Torus: N = 2 2n bisection width = 2  2 n channel width = w d bisection bandwidth = 2 n+1  w d 2 n+1  w d = 2 2n-1  w h w d / w h = 2 2n-1 / 2 n+1 = 2 n-2 = 2 n /4 =  N /4 3-D Torus: w 3d / w h = (N) 1/3 /4

6 Network Topology Graph Model: ProcessorsNodes Wires joining processorsLink

7 Linear Array 1 # of processors = n # of links = n-1 Connection (i, i+1)for i=1,2, …, n-1 Diameter(Longest Path Length)= n-1 Communication delay worst case: diameter Average case: 1/n 2   d ij 32n.....

8 Hypercube Hamming Distance D H (x, y) = # of positions in which x & y differ x & y are binary vectors D H (1100, 0111) = 3 Nodes are labeled as n-bit binary Two nodes, x & y, are adjacent if D H (x, y) =

9 Product of graphs G 1 = (V 1, E 1 ) & G 2 = (V 2, E 2 ) G 1  G 2 = G = (V, E)= (V 1  V 2, E) Two nodes in G, ((v 1, v 2 ), (v 3, v 4 )) are adjacent 1) if v 1 = v 3 and (v 2, v 4 )  E 2 or 2) if v 2 = v 4 and (v 1, v 3 )  E 1 Q 2  Q n-1 = Q n or Q k  Q n-k = Q k

10 Routing: A  B A=(a n-1, a n-2, …, a 1, a 0 ) B =(b n-1, b n-2, …, b 1, b 0 ) a n-1, a n-2, …, a 2, a 1, a 0 a n-1, a n-2, …, a 2, a 1, b 0 a n-1, a n-2, …, a 2, b 1, b 0 a n-1, a n-2, …, b 2, b 1, b 0 a n-1, b n-2, …, b 2, b 1, b 0 b n-1, b n-2, …, b 2, b 1, b 0 # of steps required = D H (A, B) Diameter of the network = n = log 2 N Average comm. delay = 1/N (( )+2( ) … n( )) = n2 n-1 /N = n2 n-1 /2 n = n/2 n1n1 n2n2n

11 Connectivity Let D H (x, y) = d n : node disjoint paths between x & y d : paths have length d n-d : paths have length d+2 Example:  , d(110000,111111)=4

Fault Diameter = n+1

13 De Bruijn Network # of nodes, N = 2 n Node A=(a n-1, a n-2, …, a 0 ) is adjacent to 1. a n-2, a n-3, …, a 1, a 0, 0 2. a n-2, a n-3, …, a 1, a 0, a n-1, a n-2, …, a a n-1, a n-2, …, a 1

14 De Bruijn Network N=2 3 = 8 2 n - 4 nodes have degree 4 2 nodes have degree 3 2 nodes have degree 2 # of links  4  2 n / 2 = 2 n

15 Routing: A  B A=(a n-1, a n-2, …, a 1, a 0 ) B =(b n-1, b n-2, …, b 1, b 0 ) a n-1, a n-2, …, a 2, a 1, a 0 a n-2, a n-3, …, a 1, a 0, b n-1 a n-3, a n-4, …, a 0, b n-1, b n-2 a n-4, a n-5, …, b n-1, b n-2, b n-3 b n-1, b n-2, …, b 2, b 1, b 0  log 2 N Example A = 1010 B =

16 a n-1, a n-2, …, a 1, a 0 a n-2, …, a 1, a 0, 0 a n-3, …, a 0, 0, 0 … 0, 0, 0, …, 0, 0 b 0, 0, 0, …, 0, 0 b 1, b 0, 0, …, 0, 0 … b n-2, b n-3, …, b 0, 0 a n-2, …, a 1, a 0, 1 a n-3, …, a 0, 1, 1 … 1, 1, 1, …, 1, 1 b 0, 1, 1, …, 1, 1 b 1, b 0, 1, …, 1, 1 … b n-2, b n-3, …, b 0, 1 b n-1, b n-2, …, b 1, b 0 2-disjoint paths

17 k-ary n-cube Multidimensional torus

18 k-ary n-cube Given a node (a n-1, a n-2, …, a 1, a 0 ) adjacent to 2n nodes given by (a n-1, a n-2, …, a 1, a 0 ±1) (a n-1, a n-2, …, a 1 ±1, a 0 ) : (a n-1 ±1, a n-2, …, a 1, a 0 ) k=8, n=3node (3 2 4) is adjacent to (3, 2, 5), (3, 2, 3) (3, 3, 4), (3, 1, 4) (4, 2, 4), (2, 2, 4) N = # of nodes in k-ary n-cube = k n

k=4, n=2 k=2, n=3 Hypercube

20 0 k=8 Lee Distance D L ((a 3, a 2, a 1, a 0 ), (b 3, b 2, b 1, b 0 )) =  min (a i - b i, b i - a i ) mod k D L (1 2 3, 3 2 1) = min (1-3, 3-1) + min (2-2, 2-2) + min (3-1, 1-3) = =

21 Routing: A  B A=(a n-1, a n-2, …, a 1, a 0 ) B =(b n-1, b n-2, …, b 1, b 0 ) a n-1, a n-2, …, a 2, a 1, a 0 ±1 a n-1, a n-2, …, a 2, a 1, a 0 ±2 : a n-1, a n-2, …, a 2, a 1, b 0 a n-1, a n-2, …, b 2, a 1 ±1, b 0 : a n-1, a n-2, …, a 2, b 1, b 0 : b n-1, b n-2, …, b 2, b 1, b 0 Example (k=5) (4,4,2,1)  (2,1,4,4) (4,4,2,1) (4,4,2,0) (4,4,2,4) (4,4,3,4) (4,4,4,4,) (4,0,4,4) (4,1,4,4) (3,1,4,4) (2,1,4,4) # of steps = D L (A, B)

22 Torus of size k n-1, k n-2, …, k 1, k 0 Mixed radix number system Processor (a n-1, a n-2, …, a 1, a 0 ) 0  a i < k i-1 i = 0, 1, 2, … n-1 Decimal value = a n-1 (k n-2 k n-3, …, k 0 ) + a n-2 (k n-3 k n-4, …, k 0 ) + …+ a 0 Two nodes, A=(a n-1, a n-2, …, a 1, a 0 ) & B =(b n-1, b n-2, …, b 1, b 0 ) are adjacent if D L (A, B) = 1

23 Torus of size k n-1, k n-2, …, k 1, k 0 Example: (8  5  4) (3,2,1) = 3  (5  4) + 2  = = Decimal to mixed radix5 17 … 1 69 = (3,2,1) 3… 2 Example: (8  8  8) (3,2,4) = 3   = 3  = … = (3,2,4) 3… 2 Radix k-number (a n-1, a n-2, …, a 1, a 0 )= a n-1 k n-1 + a n-2 k n-2 + … + a 0

24 Embedding Cycles Hypercube- Gray codes …

25 Mapping (Binary to Gray) 000         100 f(x n-1 x n-2 … x 0 ) = (g n-1 g n-2 … g 0 ) where g n-1 = x n-1 g i = x i  x i+1 i= n-2, n-3, …,  11101

26 k-ary n-cube f: Radix  Gray f(x n-1 x n-2 … x 0 ) = (g n-1 g n-2 … g 0 ) g n-1 = x n-1 g i = x i - x i+1 mod k for i= n-2, n-3, …,0

27 3-ary 2-cube RadixGray 0 00         22k=  

28 De Bruijn Network node s j+3 + s j+1 + s j = 0s j+3 = s j+1 + s j mod 2 Initial conditionss 0 =1, s 1 = 0 and s 2 = 1 s 0 s 1 s 2 s 3 s 4 s 5 s 6 s 7 s 8 s

29 How to choose this different equation? Take an n-r order difference equation whose characteristic equation is a primitive polynomial of degree r. Primitive Polynomial x 2 +x+1 x 7 +x 3 +1 x 3 +x+1 x 8 +x 4 + x 3 +x 2 +1 x 4 +x+1 x 9 +x 4 +1 x 5 +x 2 +1 x 10 +x 3 +1 x 6 +x 2 +1 s j+5 = s j+2 + s j

30 Disjoint Cycle s j+3 = s j+1 + s j +1 Initial conditionss 0 =1, s 1 = 0 and s 2 = 1 s 0 s 1 s 2 s 3 s 4 s 5 s 6 s 7 s 8 s