Utilization of OpenCL for Large Graph Problems on Graphics Processing Unit

During recent years graphics processing unit (GPU) has become an inexpensive high-performance parallel computing units and general purpose computation on graphics processing unit (GPGPU) become an alternative of conventional CPU computation, which provide massive parallel computation capability to a normal computer over a system which uses only CPU. This paper investigates the performance of the algorithm for the solution of single source shortest path (SSSP) problem, all pair shortest path (APSP) problem and graph coloring problem for large graph on GPU regardless of a specific vendor. The application programming interface (API) used for programming in graphic processing unit is open compute language (OpenCL), it is a specification for heterogeneous computing. The performance of solutions on GPU is compared with the solution on CPU in term of their execution time. The result indicates the significant improvement in the performance of the solutions on GPU over solutions on CPU.