Accelerating RDBMS Operations Using GPUs

This survey paper reviews how the Graphical Processing Unit (GPU) can be leveraged by accelerating queries for the Relational Database Management System. Queries can be accelerated in many ways like optimizing primitives, using an opcode execution model, using an in memory database, kernel fusion, batching and data placement. Data trans-fer remains a significant bottleneck and thus the mitigation methods such as database compression, differential update and unified address space are reviewed