Discuss new trends and research in parallel processingSoluti

Discuss new trends and research in parallel processing.

Solution

Parallel processing is important reaserch area.Parallel computing means solving large problem with collection of elements that works together. A problem is divided in multiple parts can be solved by using multiple computers which run on multiple processors.Computer resources can be single computer with multiple resources.Until now research in parallel computing is only theoretical.Because there was no parallel hardware.But now today parallel hardware is available.One of the mjor problems in parallel programming is coordination among the sequential code segments that comprise a parallel program.It is because of the lack of a coherent model of parallel systems. This is not hardware related problem but related to structure.There are mainly two approaches in design of parallel processing.First is parallelism achieved by language in which application is written.Second is the only source of parallelism is the explicit knowledge of the application as present with the designer. First approach example is prolog language.Second approach example is POOL (Philips Object Oriented Language).Implicit parallelism has a finer grain than explicit parallelism.

There are differents programming models for parallel processing:

Common Parallel Computing Architecture:

In this non-dedicated computers which are connected through a high speed communication media to act as a parallel computer.It requires very less effort and can be built with negligible cost. Common parallel computers is divided into 3 categories: multiprocessor computer, shared memory and distributed memory computing architecture.

Shared Memory Parallel Computing Models:In this architecture a number of processors are connected to a common central memory.The data sharing is fast but processes can corrupt each others data at the same time.

Distributed Memory Parallel Computing Models:Model BSP and LogP models are the most well known models under these Models.

BSP Model:The bulk synchronization parallel model (BSP) has three components- p numbers of processors, supersteps with periodicity L and the bandwidth factor g which is defined as the ratio of computation to communication.

   LogP Model:This model consists of 4 parameters- P numbers of computers, L (latency of message passing), O (overheads involved in message passing) and g (minimum time interval between successive messages).