Scribd
Upload
57 views

Interconnection Networks and Time Shared or Common Buses

1. Time shared or common buses are used for interconnection in distributed memory architectures. 2. Common buses use arbitration algorithms like static priority, dynamic priority, or first come first served to determine which device gains access to the bus for data transfer. 3. Static priority assigns fixed priorities to devices while dynamic priority changes priorities based on recent usage to balance load.

Uploaded by

sudhnwa ghorpade
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
57 views

Interconnection Networks and Time Shared or Common Buses

1. Time shared or common buses are used for interconnection in distributed memory architectures. 2. Common buses use arbitration algorithms like static priority, dynamic priority, or first come first served to determine which device gains access to the bus for data transfer. 3. Static priority assigns fixed priorities to devices while dynamic priority changes priorities based on recent usage to balance load.

Uploaded by

sudhnwa ghorpade
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 17

Unit 6:Distributed Memory Architecture.

Point 3–Interconnection Networks-Time


shared /Common buses

Dr. Mrs Mamata Sachin Kalas


Email id:[email protected]
Interconnection networks
• Time shared /common bus.

• Crossbar switch and multiport memories.

Interconnection Networks-Time shared


08/28/22 2
common bus
Time shared common bus

• It is totally passive unit and the transfer operations are


completed by bus interface.
• Types:
1. Single Bus organization.
2. Unidirectional bus organization.
3. Multi-bus organization.

Interconnection Networks-Time shared


08/28/22 3
common bus
Single Bus organization.
• It is reliable and inexpensive.
• Drawbacks:
central point of failure.
system expansion results in increased bus contention.
• Transfer rate limited by bandwidth and speed of the single
path.

Interconnection Networks-Time shared


08/28/22 4
common bus
Unidirectional bus organization.
• Two separate unidirectional buses are used.
• In single data transfer both the buses are involved.

Interconnection Networks-Time shared


08/28/22 5
common bus
Multi-bus organization
• Multiple bidirectional buses are provided to allow
simultaneous data transfers.
• It increases complexity of the system.

Interconnection Networks-Time shared


08/28/22 6
common bus
Bus Arbitration algorithm
• Bus arbitration algorithm control access to the bus by the
competing devices.
1. The static priority algorithm
2. The fixed time slice algorithm
3. Dynamic priority algorithm
4. The first come first served algorithm
5. Independent request implementation
6. Polling implementation

Interconnection Networks-Time shared


08/28/22 7
common bus
The static priority algorithm

Interconnection Networks-Time shared


08/28/22 8
common bus
The static priority algorithm-Daisy Chaining

• When multiple devices concurrently request the use of


bus, the device with the highest priority is granted
access to it.
• In this scheme, all services effectively assinged static
priorities according to their locations along bus grant
control line.
• Request are made on common request line BRQ
• The central bus control unit propagates bus grant
signal (BGT) if the acknowledged signal indicates the
bus to be idle.

Interconnection Networks-Time shared


08/28/22 9
common bus
The fixed time slice algorithm

• The available bus bandwidth is divided into fixed


length time slices
• Time slices are sequentially offered to each device in
round robin fashion.
• All m devices are given one out of every time slices
at fixed intervals.

Interconnection Networks-Time shared


08/28/22 10
common bus
Dynamic priority algorithm

Interconnection Networks-Time shared


08/28/22 11
common bus
Dynamic priority algorithm
• Dynamic priority algorithm allow the load balancing
characteristics to be achieved without incurring the penalty of
high wait times.
• The devices are assigned unique priorities and compete to
access the bus, but the priorities are dynamically changed.
• Two algorithms are used - 1) LRU : Least recently used
2) RDC : Rotating daisy chain
• LRU gives highest priority to the requesting device that has not
used the bus for the longest interval.
• RDC generalizes the daisy chain implementation of static
priorities.
• In RDC scheme, no central controller exists. The bus grant line
is connected from the last device back to the first.
Interconnection Networks-Time shared
08/28/22 12
common bus
The First come first served algorithm

• In FCFS scheme, the requests are simply processed in


the order received.
• This scheme is symmetric because it favors no
particular processors/device on the bus.
• It gives smallest possible average wait time and
standard deviation of all wait times.

Interconnection Networks-Time shared


08/28/22 13
common bus
Polling implementation

Interconnection Networks-Time shared


08/28/22 14
common bus
Polling implementation
• In this bus grant signal (BGT) of static daisy chain
implementation is replaced by set of log2m polling lines.
• The set of poll lines is connected to each of the devices.
• On a bus request, the controller sequences through the
device addresses using poll lines.
• When a device Di requested access recognizes its address,
it raises the SACK line.
• Priority is determined by position of the device in the
polling sequence.

Interconnection Networks-Time shared


08/28/22 15
common bus
Independent request implementation

Interconnection Networks-T1616ime
08/28/22 16
shared common bus
Independent request implementation

• In this, a separate bus request BRK, and bus grant


BGT lines are connected to each device i sharing the
bus.
• This technique permits implementation of LRU,
FCFS and other allocation algorithms.

Interconnection Networks-Time shared


08/28/22 17
common bus

You might also like