The primary accusative of this implementation is to expand the use of NVIDIA Graphics Processing Units (GPUs) to accelerate the all-purpose applications outside the graphics arena. CUDA is a programming language particularly designed for parallel computation to work. Now a day, C programming is glaringly used in industries to develop general purpose applications. Normally, a C program instruction executes sequentially and do not support data parallel computation, it increases the time complexity of a program. CUDA renders C like interface, configured for programming NVIDIA GPU which supports parallel computation of different parts of same instructions on different cores of GPU. For ordinary programmers it is very sticky to write CUDA programs because it involves various irksome tasks. Today, most of the machines come with NVIDIA graphics card which contains GPU having numerous processing cores. It is mainly used during execution of gaming, graphics and image processing kind of applications. It remains otiose during execution of general-purpose applications which results into surplus time. To properly employ the potential of available GPU cores on graphics cards for accelerating execution of applications outside graphics domain, the system implemented here provides an automatic tool that converts the directive based sequential C program and generates equivalent parallel CUDA program which will significantly enhance the speed of execution of program with help of parallel processing support. The C programmers can use this tool to enhance the speed of execution of their applications by transforming their directive based C code to CUDA C code. This tool provides simple user interface and helps to enhance the performance of the system.

Parallel Computing Serial Computing CUDA GPU HPC