Survey on Query Optimization of GPU database

In the era of artificial intelligence, CPU-driven database is not applicable in terms of data access speed and processing speed. With ultra-high-speed computing power and ultra-large data processing bandwidth capabilities, GPU is favored by people. The database branch with GPU computing as the core (GDBMS) is booming. Because of its high characteristics of concurrency (simple) operation, high throughput operation, computationally intensive (ALU unit), the ability to interact with artificial intelligence, time-space data analysis, and data visualization, it has completely changed the traditional database format. The query operation is a common operation of the database system, and the optimization of the query is always a core task. This article first systematically introduces the history of database development, and then introduces several query optimization methods that currently exist. Finally, according to the existing query optimization methods of the database, the development trend of GPU database query optimization is prospected.