Developing Library for Transport Layer of Internet Protocol Suite on CUDA platform

Abstract

Presently, the computational power of graphics processing units (GPUs) has turned them into attractive platforms for general-purpose applications at significant speed using CUDA. Compute Unified Device Architecture (CUDA) programmed, Graphic Processing Units (GPU) is rapidly becoming a major choice in high performance computing (HPC). Hence, the number of applications ported to the CUDA platform is growing high. So, a major challenge in today’s embedded world is high performance computing and to attain high precision and real time performance-which is difficult to achieve even with the most powerful CPU. In the networking world, Packet parsing is a complex task due to bit wise operation. So we can offload packet parsing task on the CUDA enable GPU. For this purpose, we are choosing to build networking library prototype, to boost the processing speed of networks on CUDA compatible GPUs. In response, we propose to develop the libraries for parsing transport layer of internet protocols on NVIDIA CUDA parallel processing platform (NVIDIA CUDA enabled GPU). With this, we can offload the protocol parsing task of Intel CPU, optimize the CPU usage and increase the performance efficiencies.
DOI: 10.17762/ijritcc2321-8169.150532