Comparing GPU and CPU in OLAP Cubes Creation

GPGPU (General Purpose Graphical Processing Unit) programming is receiving more attention recently because of enormous computations speed up offered by this technology. GPGPU is applied in many branches of science and industry not excluding databases, even if this is not the primary field of expected benefits. In this paper a typical time consuming database algorithm, i.e. OLAP cube creation, implemented on GPU is compared to its CPU counterpart by analysis of performance, scalability, programming and optimisation ease. Results are discussed formulating roadmap for future GPGPU applications in databases.