| Abstract : | Join is an expensive but important database operation. This paper examines how the latest computer architectures and CPU performance features improve one of the fastest join algorithms up to now – Radix Cluster Hash Join – and compares it to some traditional join algorithms like hash join and nested loop join. We start by implementing two variations of the Radix Cluster Hash Join algorithm and then we conduct several experiments, using one, four and eight threads. We keep optimizing the parameters of each algorithm in order to obtain the best possible performance and compare it to the traditional algorithms in order to illustrate the improved performance.
|
|---|
Copyright © 2013 Library AUEB. All rights reserved.