Despite the simplicity brought by the utilisationof a uniprocessor and the advantage of having astraightforward hardware not forgetting its simplesoftware. Yet it doesn’t satisfy many uprising applicationswith their need for a processor with high performance.This is where multiprocessors emerge as a solution forsuch demanding applications where the need for morefunctionality.
The use of a multiprocessor comes with theadvantage of minimising processing time, minimisinglatency and most important is the given high performance.The gained advantages are tightly connected withsatisfying a very important feature in the multiprocessingenvironment which is load balance. Typically, loadbalancing can be achieved when tasks are evenlyscheduled (distributed) across the connected processingelements.
Through the discussion of MPSoC environmentthere are many controversial concepts that would possiblycreate some kind of confusion in dealing with them.Concepts like Mechanism and Policy where the firstprovides the power to execute an action while policydetermines what to do with the mechanism. Some timesthere is a difficulty in distinguishing between parallel anddistributed systems, which will be distinguished accordingto the autonomy of individual node in the system.
Autonomy is confirmed in distributed systems, yet this isnot the case in parallel systems. When a node has thefreedom to behave differently than the rest of the nodes inthe system, it is considered an autonomous node. Designautonomy, communication autonomy, execution autonomyand administrative autonomy are the four components tothe autonomy of amultiprocessor system. Earlierdistinctions were grounded on the possibility of needinginterrupt to access some parts of memory like ifcommunication among processors is through sharedmemory (tight coupling), or through message passing(loose coupling).The mix in the concept between a job and a taskhas always been faced in multiprocessing environment.The computing unit in a computing systems is called atask while various tasks functioning to achieve a commongoal are known as a job. So in such environments thehanded situation is having various jobs each composed ofmany tasks, and all competing to be served by a processor.
The allocation of tasks on a processor is either assigningseveral processors to a single job which is called spacesharing or assigning several tasks to a single processor andthat’s called time sharing.In the MPSoC field, the recently reconfigurablemultiprocessor which is mostly FPGA-basedmultiprocessor is a modern and growingly significanttrend. Reconfigurable multiprocessor brings the vantagesof rapidly facilitating prototype and permitting study intomore efficient architectures and communicationstechniques. These multiprocessors also exclude thedrawbacks of MPSoC ASIC production. ReconfigurableMultiprocessor Systems, or often known softmultiprocessor is mainly proposed as a meanto help inprototyping systems for later implementation on an ASIC.At the present, FPGAs opened new horizons not just forimplementing prototypes, but for the benefit of the final VOL. 12, NO. 1, JANUARY 2017 ISSN 1819-6608ARPN Journal of Engineering and Applied Sciences©2006-2017 Asian Research Publishing Network (ARPN).
All rights reserved.www.arpnjournals.
com 2designs as well. The development of FPGA capacitypermits designers to implement an entire multiprocessorsystem on one FPGA. The FPGA chief companies providethe possibility of utilising soft-core processors particularlydesigned to well match in the FPGA. At the same time,FPGAs permit the utilisation of hard-core processors.Moreover, FPGAs are fitted with on-chip memory blocks,peripherals, and interconnection circuitry.
One of theFPGA based multiprocessor strengths is the run-timereconfigurability. This characteristic providesmultiprocessor systems an adaptation feature to aparticular application, making the designed system to earnmuch of flexibility.The power provided by the utilisation ofreconfigurable processor has been examined by 1-3. Theresearchers did confirm that reconfigurable multiprocessorbased FPGA provided adaptations for many facedchallenges.Since the second part of this paper will deal withload balancing the confusion between load balancing andload sharing should be cleared. Load sharing permitsprocessors which are busy to offload some of their loads(work) to idle or less busy processors. Load balancing is aparticular case of load sharing, aims to keep the load even(or balanced) across all processors and this is the goal ofglobal scheduling algorithms.
The threshold load is animportant term in load balancing strategies. The totalloadto a processor that every load would come further to tha