Increasing Performances of Sca Applications That Use Opencl

The Open Computing Language (OpenCL) can be used in conjunction with the Software Communications Architecture (SCA) to build very portable applications that execute across heterogeneous platforms consisting of General Purpose Processors (GPPs), Digital Signal Processors (DSPs), Field Programming Gate Arrays (FPGAs), and Graphics Processing Units (GPUs). This paper starts with an overview of how SCA components can be built using OpenCL. It compares performance metrics of an application implemented as several OpenCL-SCA components with the metrics of a variation of the application that uses a reducedcopy technique for the data being processed. The paper provides a detailed discussion on how the data flows in and out of the OpenCL device memory as it travels from a component to the next. The paper describes a novel approach that minimizes the number of copies made as the data flows through the different signal processing components of an SCA application. The paper concludes by identifying further research topics that could be investigated on this subject.