Sorting very large text data in multi GPUs

Text data sorting is used in many fields and plays an important role in defining the overall speed and performance. For example, text sort is used in finance to update stock market data. Text data processing consumes huge amounts of memory for storage and computation purposes. Hence we propose a technique to sort huge text data which combines the Radix sort algorithm and the Bucket sort algorithm to obtain a scalable, parallel, and efficient sorting method which can be applied on multiple Graphic Processor Unit (GPU) cards. The basic idea is to divide the text data into multiple buckets and then sort individual buckets in different GPUs in parallel eventually merging the sorted bucket into a final sorted result. On individual GPUs, we will implement Radix sort which will convert text (ASCII) into integers and perform sorting using this algorithm.