GPU in applications with non-relational DBs

Rapid speeds have been acquired throughout the past decade using GPUs and NoSQL-databases. This paper presents a concept of a GPU-extended non-relational database management system. The research focuses on implementing kernels and performing the basic aggregation functions over a JSON file. Different comparisons with the Numpy library, a CPU-counterpart and MongoDb show the importance of the concept. The hypothesis is to check if GPU can speed up NoSQL-database queries which we prove using our methods.