This Thesis is about the study of the OpenCL. It is considered as a framework where programs which are written can be executed on multiple platforms which consist of several units like central processing unit, processors, and also graphical processing unit.
OpenCL is capable of performing the computations in a parallel manner. The parallel execution can be either task based or even can be data based. The main aim of this project is to study the new platform OpenCL which is launched in recently.
There is a huge development from the past years in the field of the High Scale Computation especially much development is observed in the field of the General Purpose Graphical processing unit.
With the feature of the platform independency OpenCL is demanded by many when compared to that of technologies of CUDA. Which are capable of performing the computations on fixed number of graphical card.
When we compare the non CUDA, then we have only option of the heterogeneous system called OpenCL for the sake of the high scalable feature and when we speak about CUDA we have options either OpenCL or even CUDA.
In this project a comparison is done based on the performance of the two systems which are CUDA based systems. Based on the performance we will reach to the conclusions which will more suites for the demand of high scale computation.
We can conclude that OpenCL is capable of performing the computations in a parallel manner. The parallel execution can be either task based or even can be data based.
Download Analysis of OpenCL on CUDA Platform IIT Thesis for CSE Students.