Reinforced Learning Based Spatial Sorting Based Dynamic Task Allocation on Networked Multicore GPU Processors

Abstract

Owing to the advancements in cloud computing our lives are significantly altering the means of utilizing data by the data intensive business and research. The coming era of ubiquitous computing is greatly supported by evolution of cloud computing with networked multicore GPU processers to avail consistent data utilization. In such a domain, computing, data stockpiling and correspondence turns into a utility. In this paper, a new algorithm is devised that aids in ranking the tasks based on generated cluster indices. The proposed algorithm offers a new strategy for simultaneous task partitioning, its ranking and load assignment, thus improving the computational performance of the given workload.