Database scan variants on modern CPUs: A performance study

Abstract

Main-memory databases rely on highly tuned database operations to achieve peak performance. Recently, it has been shown that different code optimizations for database operations favor different processors. However, it is still not clear how the combination of code optimizations (e.g., loop unrolling and vectorization) will affect the performance of database algorithms on different processors. In this paper, we extend prior studies by an in-depth performance analysis of different variants of the scan operator. We find that the performance of the scan operator for different processors gets even harder to predict when multiple code optimizations are combined. Since the scan is the most simple database operator, we expect the same effects for more complex operators such as joins. Based on these results, we identify practical problems for a query processor and discuss how we can counter these challenges in future work.