登入選單

返回Google圖書搜尋

Applying Artificial Intelligence to Predict the Performance of Data-dependent Applications

Paula FritzscheDolores RexachsEmilio Luque

出版	INTECH Open Access Publisher, 2008
ISBN	95376190369789537619039
URL	http://books.google.com.hk/books?id=uMjuoAEACAAJ&hl=&source=gbs_api

註釋

This chapter introduces a general novel methodology to estimate the performance order of data-dependent parallel algorithms. It is important to understand that the parallel performance achieved depends on several factors, including the application, the parallel computer, the data distribution, and also the methods used for partitioning the application and mapping its components onto the architecture. Briefly, the general methodology works as follows. It begins by designing a certain number of instances and collecting their execution-time data. A well-designed instance guides the experimenters in choosing what experiments actually need to be performed in order to provide a representative sample. A data-mining process then explores these collected data in search of patterns and/or relationships detecting the main parameters that affect performance. These common properties are modelled numerically so as to generate an analytical formulation of the execution time. The methodology views the algorithm being study as a black box in which the measured values for this limited number of inputs arrive, are processed, and then produce a multiple-linear-regression model. Finally, the regression equation allows for predicting how the algorithm will perform when given new input data sets.