A Critical Performance Study of Memory Mapping on Multi-Core Processors: An Experiment with k-means Algorithm with Large Data Mining Data Sets

Abstract

Increased availability of Multi-Core processors is forcing us to redesign algorithms and applications so as to exploit the available computational power from multiple cores. It is not un-common to employ memory mapping of files in applications involving huge I/O bandwidth to improve the response/service times. This paper mainly focuses on performance of memory mapped files on MultiCore processors. Experiments are carried out with k-means algorithm, a popular Data mining (DM) clustering algorithm, to explore the potential of Multi-Core hardware under OpenMP API and POSIX threads. Observations are made both with static and dynamic threads of OpenMP. Experiments are also conducted with both simulated and real data sets. Experiments indicate that memory mapping of files gives considerable benefit on MultiCore processors also. In addition, the benefit increased with increased physical memory size. Also, the benefit of memory mapping with the selected algorithm is increasing with number of cores.

M-FDBSCAN: A multicore density-based uncertain data clustering algorithm

“…Our proposed algorithm is one of the first on adapting data mining algorithms for multicore system architecture. In [5], a k-means clustering algorithm for multicore system architecture was presented, but the algorithm was based on certain data.…”

“…After the k-means clustering algorithm was applied to the relevant part of the dataset by each dedicated core, simultaneously, a merge operation was run to get the final clusters. In [5], the dataset was composed of certain data.…”

“…In [5], one of the most well-known clustering algorithms, the k-means algorithm, was reimplemented for multicore system architecture. To each core, some part of the dataset was assigned for clustering.…”

“…In [5], a k-means clustering algorithm for multicore system architecture was presented, but the algorithm was based on certain data.…”

A collaborative divide-and-conquer K-means clustering algorithm for processing large data

“…Meanwhile, there also has been a lot of work on the parallel K-means algorithm and implementation, such as parallel implementation based on SIMD hypercube network [25,29], master/slave message passing architecture [11,13,21,22,35], shared memory multi-core processor [30,3], GPU [12] and MapReduce programming model [36,2]. Based on the probability-based seeding approach K-Means++ [6], Bahmani et al proposed a parallel seeding approach that can find a good initial set of centers rapidly [8].…”