Skip to Main Content
A shared memory parallel k-NN algorithm for M-tree index structure is introduced in this paper, which is called SMP A-NN. The processing of the pending request (PR) queue is a core operation in the traditional k-NN query algorithm, which is also time-consuming. Therefore, we separate the long queue into multi-parts and assign them to different threads. This improvement takes full advantage of SMP architecture and can keep good load balancing between threads. We implement this algorithm on a common shared memory parallel PC server. A lot of experiments have illustrated the proposed SMP k-NN can speed up the query and improve the performance of M-tree for large dataset.