SlideShare a Scribd company logo
2
Most read
8
Most read
19
Most read
PARALLEL AND DISTRIBUTED COMPUTING
INTERCONNECTION TOPOLOGIES
NETWORK TOPOLOGIES
Two types of topologies
 Shared Network: Communicate at most one message at a
time that’s why limited scalability
 Switched Network: Simultaneously transfer several messages
between different pairs of nodes.
NETWORK TOPOLOGIES
 Degree: Maximum number of neighbors of any node
 Diameter: length of longest of all shortest path
 BisectionWidth: minimum number of edges(links) to be
removed to disconnect the network into two halves. In case of
odd number of nodes one half can include one more node.
NETWORK TOPOLOGIES
 Constant Degree: Should be constant i.e. independent of
network size.This property allows scale large number of nodes
without adding excessive number of connections.
 Low Diameter: Should be minimum in order to provide
efficient communication between any pair of nodes
 High BisectionWidth: low bisection width can slow down
communication and thus limit the performance of applications.
Parallel and Distributed Computing Chapter 11
NETWORK TOPOLOGIES
 Linear Array
 2D Mesh/torus
 3D Mesh/torus
 BinaryTree
 Hypercube
LINEAR ARRAY
 Every node is connected to left and its right neighbor and thus
degree is 2 i.e. deg(Ln)=2
 Longest distance between any two nodes is between leftmost
node and the right most node i.e. Diam(Ln)=n-1
 Bisection width is one i.e. BW(Ln)=1, Since only one link
between P(n-1)/2 and P(n/2) needs to be removed in order to split
into two disconnected halves.
2D MESHTOPOLOGY
 Nodes are arranged in a (usually
square) grid
 Each node has most four neighbors i.e.
deg(M4,4)=√n=4
 Longest distance between P0,0 and P3,3
Leading to diameter of 6 i.e.
Daim(Mk,k)=2(√n-1)=6.
 Removing all links between second
and third row/column disconnects
M4,4 into equal halves.Thus bisection
width is bw(Mk,k)=(√n =4)
2D TORUSTOPOLOGY
 A frequently used extension of mesh is torus
 2D Torus Tk,k is extends Mk,k by adding wrap-around
edges to directly connect left and right node of each
raw as well the bottom and top node of each
column.
 Each node has most four neighbors i.e. deg(M4,4)=4
 Longest distance between P0,0 and P3,3 Leading to
diameter of 6 i.e. Daim(Mk,k)=(√n-1)=3.
 Removing all links between second and third
row/column disconnects M4,4 into equal halves.Thus
bisection width is bw(Mk,k)=(2√n = 8)
3D MESHTOPOLOGY
 Nodes are arranged in a (usually
square) grid
 Each node has most six neighbors
i.e. degree is Deg(Mk,k,k)=6
 Longest distance between P0,0 and
P3,3 Leading to diameter of 6 i.e.
Daim(Mk,k,k)=3(3√n-1).
 Bisection width can be calculated as
BW(Mk,k)=( n2/3)
3D MESHTOPOLOGY
 3D Mesh topology has many desirable features such as a
constant degree, relatively low diameter, and relatively high
bisection width.
 Thus it has been used as an interconnection network in top
supercomputers
3D TORUSTOPOLOGY
 Nodes are arranged in a (usually
square) grid
 Similar to 3d-Mesh topology just by
adding wrap-around edges
BINARY TREE TOPOLOGY
 In order to reduce network diameter, a tree based structure can be
used.
 Each node is connected to its parent and two Childs hence degree
is deg(BTd)=3
 The longest distance is when travelling between left most node to
right most half of tree, which requires growing up to the root and
then down again so Diam(BTd) = 2(d-1) = 2log2 (n+1).
 In-spite of constant degree and diameter is low disadvantage of
binary tree is its bisection width.
BINARY TREE TOPOLOGY
 Removing only a single link
with the root, we can split
the network into two
disconnected components
i.e. bitwise width bw(BTd)=1
HYPERCUBE NETWORK TOPOLOGY
 Hypercube network can be
represented as a graph with
n=2d nodes that has each
vertex labeled with a distinct
bit of length d
 Vertices that are connected iff
bits string differ in exactly one
bit.
Parallel and Distributed Computing Chapter 11
HYPERCUBE NETWORK TOPOLOGY
 Each node is connected to d
other nodes so degree will
deg(Qd)=d=log2(n)
 Longest distance between two
nodes diam(Qd)=d=log2(n)
 Removing all links between nodes
starting with label O and all
nodes starting with 1
disconnects Hd into two halves
i.e. it holds bw(Qd)=n/2
NETWORK TOPOLOGY
HYPERCUBE NETWORK TOPOLOGY
 Overall hypercube has the highest bisection width
 Furthermore diameter is very low (logarithmic)
 But disadvantage is non constant degree. i.e. number of required
links per node is logarithmic
 Making it difficult to scale up to large number of nodes.

More Related Content

PDF
Parallel and Distributed Computing Chapter 10
PPTX
Lecture 2 - Asynchrnous and Synchronous Computation & Communication.pptx
PDF
Parallel and Distributed Computing Chapter 12
PPT
File replication
PDF
Interconnection Network
PDF
Parallel and Distributed Computing chapter 3
PDF
Parallel and Distributed Computing Chapter 8
PPS
Deadlock
Parallel and Distributed Computing Chapter 10
Lecture 2 - Asynchrnous and Synchronous Computation & Communication.pptx
Parallel and Distributed Computing Chapter 12
File replication
Interconnection Network
Parallel and Distributed Computing chapter 3
Parallel and Distributed Computing Chapter 8
Deadlock

What's hot (20)

PPTX
SYNCHRONIZATION IN MULTIPROCESSING
PDF
Basic communication operations - One to all Broadcast
PPTX
Distributed and clustered systems
PPTX
Design Goals of Distributed System
PPT
Clock synchronization in distributed system
PDF
Distributed deadlock
PPTX
Load Balancing In Distributed Computing
PPT
Deadlock
PPTX
5.2.2. Memory Consistency Models.pptx
PPT
recursive transition_networks
PPT
parallel programming models
PDF
Parallel programming model, language and compiler in ACA.
PPTX
Kernels and its types
PPTX
Design Issues of Distributed System (1).pptx
PDF
Chapter 5: Mapping and Scheduling
PPTX
RPC: Remote procedure call
DOC
Distributed Mutual exclusion algorithms
PPT
distributed shared memory
PDF
VTU 6TH SEM CSE COMPUTER NETWORKS 2 SOLVED PAPERS OF JUNE-2013 JUNE-14 & JUNE...
PDF
Distributed Operating System_4
SYNCHRONIZATION IN MULTIPROCESSING
Basic communication operations - One to all Broadcast
Distributed and clustered systems
Design Goals of Distributed System
Clock synchronization in distributed system
Distributed deadlock
Load Balancing In Distributed Computing
Deadlock
5.2.2. Memory Consistency Models.pptx
recursive transition_networks
parallel programming models
Parallel programming model, language and compiler in ACA.
Kernels and its types
Design Issues of Distributed System (1).pptx
Chapter 5: Mapping and Scheduling
RPC: Remote procedure call
Distributed Mutual exclusion algorithms
distributed shared memory
VTU 6TH SEM CSE COMPUTER NETWORKS 2 SOLVED PAPERS OF JUNE-2013 JUNE-14 & JUNE...
Distributed Operating System_4
Ad

Similar to Parallel and Distributed Computing Chapter 11 (20)

PPTX
presentation on Topology
PDF
A Survey on Area Planning for Heterogeneous Networks
PPTX
A Study on data analysis in Oncology.pptx
PPTX
A Study on data analysis in Oncology.pptx
PPT
1414407496.309801 network topologies (1)
PPT
1414407496.309801 network topologies
PPT
01 network topologies
PPT
1414407496.309801 network topologies
PPTX
Mesh topology
PDF
5113jgraph02
PPTX
Topology
PPTX
Topology And Mesh Topology Group Presentation
PDF
Mesh topology
PPTX
NETWORK TOPOLOGY
PPTX
Network topology
PPTX
Networking.pptx
PDF
Network Topology's
PPTX
Physical organization of parallel platforms
PPTX
Topology
PPT
mathematics of network science: basic definitions
presentation on Topology
A Survey on Area Planning for Heterogeneous Networks
A Study on data analysis in Oncology.pptx
A Study on data analysis in Oncology.pptx
1414407496.309801 network topologies (1)
1414407496.309801 network topologies
01 network topologies
1414407496.309801 network topologies
Mesh topology
5113jgraph02
Topology
Topology And Mesh Topology Group Presentation
Mesh topology
NETWORK TOPOLOGY
Network topology
Networking.pptx
Network Topology's
Physical organization of parallel platforms
Topology
mathematics of network science: basic definitions
Ad

More from AbdullahMunir32 (16)

PDF
Mobile Application Development-Lecture 15 & 16.pdf
PDF
Mobile Application Development-Lecture 13 & 14.pdf
PDF
Mobile Application Development -Lecture 11 & 12.pdf
PDF
Mobile Application Development -Lecture 09 & 10.pdf
PDF
Mobile Application Development -Lecture 07 & 08.pdf
PDF
Mobile Application Development Lecture 05 & 06.pdf
PDF
Mobile Application Development-Lecture 03 & 04.pdf
PDF
Mobile Application Development-Lecture 01 & 02.pdf
PDF
Parallel and Distributed Computing Chapter 13
PDF
Parallel and Distributed Computing Chapter 9
PDF
Parallel and Distributed Computing Chapter 7
PDF
Parallel and Distributed Computing Chapter 6
PDF
Parallel and Distributed Computing Chapter 5
PDF
Parallel and Distributed Computing Chapter 4
PDF
Parallel and Distributed Computing Chapter 2
PDF
Parallel and Distributed Computing chapter 1
Mobile Application Development-Lecture 15 & 16.pdf
Mobile Application Development-Lecture 13 & 14.pdf
Mobile Application Development -Lecture 11 & 12.pdf
Mobile Application Development -Lecture 09 & 10.pdf
Mobile Application Development -Lecture 07 & 08.pdf
Mobile Application Development Lecture 05 & 06.pdf
Mobile Application Development-Lecture 03 & 04.pdf
Mobile Application Development-Lecture 01 & 02.pdf
Parallel and Distributed Computing Chapter 13
Parallel and Distributed Computing Chapter 9
Parallel and Distributed Computing Chapter 7
Parallel and Distributed Computing Chapter 6
Parallel and Distributed Computing Chapter 5
Parallel and Distributed Computing Chapter 4
Parallel and Distributed Computing Chapter 2
Parallel and Distributed Computing chapter 1

Recently uploaded (20)

PDF
A GUIDE TO GENETICS FOR UNDERGRADUATE MEDICAL STUDENTS
PDF
3rd Neelam Sanjeevareddy Memorial Lecture.pdf
PPTX
human mycosis Human fungal infections are called human mycosis..pptx
PDF
The Lost Whites of Pakistan by Jahanzaib Mughal.pdf
PPTX
GDM (1) (1).pptx small presentation for students
PDF
RMMM.pdf make it easy to upload and study
PDF
O7-L3 Supply Chain Operations - ICLT Program
PPTX
Final Presentation General Medicine 03-08-2024.pptx
PPTX
school management -TNTEU- B.Ed., Semester II Unit 1.pptx
PDF
grade 11-chemistry_fetena_net_5883.pdf teacher guide for all student
PPTX
Tissue processing ( HISTOPATHOLOGICAL TECHNIQUE
PPTX
Institutional Correction lecture only . . .
PPTX
Presentation on HIE in infants and its manifestations
PDF
O5-L3 Freight Transport Ops (International) V1.pdf
PDF
A systematic review of self-coping strategies used by university students to ...
PDF
GENETICS IN BIOLOGY IN SECONDARY LEVEL FORM 3
PPTX
Pharma ospi slides which help in ospi learning
PDF
Computing-Curriculum for Schools in Ghana
PDF
Microbial disease of the cardiovascular and lymphatic systems
PPTX
master seminar digital applications in india
A GUIDE TO GENETICS FOR UNDERGRADUATE MEDICAL STUDENTS
3rd Neelam Sanjeevareddy Memorial Lecture.pdf
human mycosis Human fungal infections are called human mycosis..pptx
The Lost Whites of Pakistan by Jahanzaib Mughal.pdf
GDM (1) (1).pptx small presentation for students
RMMM.pdf make it easy to upload and study
O7-L3 Supply Chain Operations - ICLT Program
Final Presentation General Medicine 03-08-2024.pptx
school management -TNTEU- B.Ed., Semester II Unit 1.pptx
grade 11-chemistry_fetena_net_5883.pdf teacher guide for all student
Tissue processing ( HISTOPATHOLOGICAL TECHNIQUE
Institutional Correction lecture only . . .
Presentation on HIE in infants and its manifestations
O5-L3 Freight Transport Ops (International) V1.pdf
A systematic review of self-coping strategies used by university students to ...
GENETICS IN BIOLOGY IN SECONDARY LEVEL FORM 3
Pharma ospi slides which help in ospi learning
Computing-Curriculum for Schools in Ghana
Microbial disease of the cardiovascular and lymphatic systems
master seminar digital applications in india

Parallel and Distributed Computing Chapter 11

  • 1. PARALLEL AND DISTRIBUTED COMPUTING INTERCONNECTION TOPOLOGIES
  • 2. NETWORK TOPOLOGIES Two types of topologies  Shared Network: Communicate at most one message at a time that’s why limited scalability  Switched Network: Simultaneously transfer several messages between different pairs of nodes.
  • 3. NETWORK TOPOLOGIES  Degree: Maximum number of neighbors of any node  Diameter: length of longest of all shortest path  BisectionWidth: minimum number of edges(links) to be removed to disconnect the network into two halves. In case of odd number of nodes one half can include one more node.
  • 4. NETWORK TOPOLOGIES  Constant Degree: Should be constant i.e. independent of network size.This property allows scale large number of nodes without adding excessive number of connections.  Low Diameter: Should be minimum in order to provide efficient communication between any pair of nodes  High BisectionWidth: low bisection width can slow down communication and thus limit the performance of applications.
  • 6. NETWORK TOPOLOGIES  Linear Array  2D Mesh/torus  3D Mesh/torus  BinaryTree  Hypercube
  • 7. LINEAR ARRAY  Every node is connected to left and its right neighbor and thus degree is 2 i.e. deg(Ln)=2  Longest distance between any two nodes is between leftmost node and the right most node i.e. Diam(Ln)=n-1  Bisection width is one i.e. BW(Ln)=1, Since only one link between P(n-1)/2 and P(n/2) needs to be removed in order to split into two disconnected halves.
  • 8. 2D MESHTOPOLOGY  Nodes are arranged in a (usually square) grid  Each node has most four neighbors i.e. deg(M4,4)=√n=4  Longest distance between P0,0 and P3,3 Leading to diameter of 6 i.e. Daim(Mk,k)=2(√n-1)=6.  Removing all links between second and third row/column disconnects M4,4 into equal halves.Thus bisection width is bw(Mk,k)=(√n =4)
  • 9. 2D TORUSTOPOLOGY  A frequently used extension of mesh is torus  2D Torus Tk,k is extends Mk,k by adding wrap-around edges to directly connect left and right node of each raw as well the bottom and top node of each column.  Each node has most four neighbors i.e. deg(M4,4)=4  Longest distance between P0,0 and P3,3 Leading to diameter of 6 i.e. Daim(Mk,k)=(√n-1)=3.  Removing all links between second and third row/column disconnects M4,4 into equal halves.Thus bisection width is bw(Mk,k)=(2√n = 8)
  • 10. 3D MESHTOPOLOGY  Nodes are arranged in a (usually square) grid  Each node has most six neighbors i.e. degree is Deg(Mk,k,k)=6  Longest distance between P0,0 and P3,3 Leading to diameter of 6 i.e. Daim(Mk,k,k)=3(3√n-1).  Bisection width can be calculated as BW(Mk,k)=( n2/3)
  • 11. 3D MESHTOPOLOGY  3D Mesh topology has many desirable features such as a constant degree, relatively low diameter, and relatively high bisection width.  Thus it has been used as an interconnection network in top supercomputers
  • 12. 3D TORUSTOPOLOGY  Nodes are arranged in a (usually square) grid  Similar to 3d-Mesh topology just by adding wrap-around edges
  • 13. BINARY TREE TOPOLOGY  In order to reduce network diameter, a tree based structure can be used.  Each node is connected to its parent and two Childs hence degree is deg(BTd)=3  The longest distance is when travelling between left most node to right most half of tree, which requires growing up to the root and then down again so Diam(BTd) = 2(d-1) = 2log2 (n+1).  In-spite of constant degree and diameter is low disadvantage of binary tree is its bisection width.
  • 14. BINARY TREE TOPOLOGY  Removing only a single link with the root, we can split the network into two disconnected components i.e. bitwise width bw(BTd)=1
  • 15. HYPERCUBE NETWORK TOPOLOGY  Hypercube network can be represented as a graph with n=2d nodes that has each vertex labeled with a distinct bit of length d  Vertices that are connected iff bits string differ in exactly one bit.
  • 17. HYPERCUBE NETWORK TOPOLOGY  Each node is connected to d other nodes so degree will deg(Qd)=d=log2(n)  Longest distance between two nodes diam(Qd)=d=log2(n)  Removing all links between nodes starting with label O and all nodes starting with 1 disconnects Hd into two halves i.e. it holds bw(Qd)=n/2
  • 19. HYPERCUBE NETWORK TOPOLOGY  Overall hypercube has the highest bisection width  Furthermore diameter is very low (logarithmic)  But disadvantage is non constant degree. i.e. number of required links per node is logarithmic  Making it difficult to scale up to large number of nodes.