0% found this document useful (0 votes)
16 views32 pages

Lecture 3 Network Topologies

The document discusses various topics related to parallel processing network topologies. It begins by defining different network topologies including bus, ring, star, and hybrid topologies. It then describes the omega switching network, which uses a self-routing algorithm and perfect shuffles to connect processors and memory in a scalable network. The document provides an example of how self-routing works in an omega network and discusses the path contention problem that can occur when multiple processors attempt to access memory modules using the same network paths.

Uploaded by

wardabibi69
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
16 views32 pages

Lecture 3 Network Topologies

The document discusses various topics related to parallel processing network topologies. It begins by defining different network topologies including bus, ring, star, and hybrid topologies. It then describes the omega switching network, which uses a self-routing algorithm and perfect shuffles to connect processors and memory in a scalable network. The document provides an example of how self-routing works in an omega network and discusses the path contention problem that can occur when multiple processors attempt to access memory modules using the same network paths.

Uploaded by

wardabibi69
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 32

Parallel Processing

Network Topologies
Nazia Shahzadi

Faculty of Computer Science and


Engineering

[email protected]
Topics for Today

Interconnection networks
• Topologies
• Characteristics
—cost
• —performance
• Omega Switching
Network

2
Metrics for Interconnection Networks

• Degree
—number of links per node
• Diameter
—longest distance between two nodes in the network
• Cost:
—~ # links and switches

3
Network Topologies

• Many network topologies


• Tradeoff: performance vs. cost
• Machines often implement
hybrids of multiple topologies
—why?
– Fault tolerant
– cost
– available components

4
Network Topology

Network topology is the arrangement of various computer network elements


(links, nodes, etc.).
Topology = Layout
Network topology defines a network's layout or structure.
 There are two types of network topologies:
 Physical Topology
 Logical Topology
Physical and Logical Topology
Physical Topology:
 Physical topology indicates the arrangement of different elements
of a network. It reflects the physical layout of devices and cables
to form a connected network. It is concerned with the essentials of
the network ignoring minute details like transfer of data and device
type.
Logical Topology:
 Logical Topology reflects the arrangement of devices and their
communication. It is the transmission of data over physical
topology. It is independent of physical topology, irrespective of the
arrangements of nodes.
Network Topologies: Bus
 Bus Topology is a network type in which every computer and network device is
connected to a single cable. It is bi-directional. It is a multi-point connection and a
non-robust topology because if the backbone fails the topology crashes.

 Bus topology is a network topology used in the context of parallel


processing to connect multiple processing nodes or devices in a linear fashion.
While bus topology is not the most common choice for high-performance
parallel processing
7
Ring Topology
In a Ring Topology, it forms a ring connecting devices with exactly two
neighboring devices. A number of repeaters are used for Ring topology with a large
number of nodes, because if someone wants to send some data to the last node in the
ring topology with 100 nodes, then the data will have to pass through 99 nodes to
reach the 100th node.

 In a ring topology, data can be passed sequentially from one node to the next
until it reaches its destination. This sequential data passing can be advantageous
in certain parallel processing applications, such as tasks that involve
iterative computations
Star Topology

In Star Topology, all the devices are connected to a single hub through a cable. This
hub is the central node and all other nodes are connected to the central node.

 Each processing node in a star topology is connected directly to the central hub
or switch. This isolation can be useful when there is a need to physically or
logically isolate nodes from one another. For example, in secure parallel
processing environments, isolating nodes can enhance security by
preventing unauthorized access between nodes.
Hybrid Topology
A hybrid topology is a type of network topology that uses two or more differing
network topologies. These topologies can include a mix of bus topology, mesh
topology, ring topology, star topology, and tree topology. The choice to use a hybrid
topology over a standard topology depends on the needs of a business, school, or the
users. The number of computers, their location, and desired network performance
are all factors in the decision.

 Hybrid topology provides the flexibility to customize the network infrastructure


to meet the specific requirements of the parallel processing application. It
allows you to adapt the topology to the unique communication and
computational needs of the system.
Omega Switching Network

• An Omega network is used to connect processors and memory in a


large scalable network.

Consider an 8*8 Omega switching network


and a 2*2 switch size
No of Inputs = 8
No of Outputs = 8
No of stages = 3
No of switches in a stage = no of inputs/switch size
= 8/2
=4

11
Omega Network Switches

• 2×2 switches connect perfect shuffles


• Each switch operates in two modes

Pass-through Cross-over

12
Omega Network
Omega network follows the concept of “Left Rotate”
Omega Switching Network
• Omega network has a self-routing property. It works for each
possible input.
• The interconnection between the stages is defined by the logic of
“Rotate left”.

14
Example of Self Routing
 If the bit is “0” then send the cell out “upper
port”
 If the bit is “1” then send the cell out “lower
port”
Suppose Processor “1” wants to access memory
module “4” that is 100
Example of Self Routing

Suppose Processor “1” wants to access memory


module “4” that is 100
Example of Self Routing

Suppose Processor “1” wants to access memory


module “4” that is 100
Example of Self Routing

Suppose Processor “1” wants to access memory


module “4” that is 100
Example of Self Routing

Suppose Processor “1” wants to access memory


module “4” that is 100
Example of Self Routing

Suppose Processor “1” wants to access memory


module “4” that is 100
Example of Self Routing

Suppose Processor “1” wants to access memory


module “4” that is 100
Example of Self Routing

Suppose Processor “1” wants to access memory


module “4” that is 100
Path Contention Problem

 Omega switching network results in performance degradation when two


processors want to access memory modules that cause problems in the path.
 It results in less throughput by the system.
Path Contention Problem
Path Contention Problem
Path Contention Problem
Path Contention Problem
Path Contention Problem
Path Contention Problem
Path Contention Problem
Path Contention Problem
End

You might also like