The patent badge is an abbreviated version of the USPTO patent document. The patent badge does contain a link to the full patent document.
The patent badge is an abbreviated version of the USPTO patent document. The patent badge covers the following: Patent number, Date patent was issued, Date patent was filed, Title of the patent, Applicant, Inventor, Assignee, Attorney firm, Primary examiner, Assistant examiner, CPCs, and Abstract. The patent badge does contain a link to the full patent document (in Adobe Acrobat format, aka pdf). To download or print any patent click here.
Patent No.:
Date of Patent:
Mar. 20, 2012
Filed:
Jul. 24, 2009
Liang Chen, Shanghai, CN;
Kuan Feng, Shanghai, CN;
Yonghua Lin, Beijing, CN;
Sheng Xu, Shanghai, CN;
Liang Chen, Shanghai, CN;
Kuan Feng, Shanghai, CN;
Yonghua Lin, Beijing, CN;
Sheng Xu, Shanghai, CN;
International Business Machines Corporation, Armonk, NY (US);
Abstract
The present invention provides a method and apparatus for partitioning, sorting a data set on a multi-processor system. Herein, the multi-processor system has at least one core processor and a plurality of accelerators. The method for partitioning a data set comprises: partitioning iteratively said data set into a plurality of buckets corresponding to different data ranges by using said plurality of accelerators in parallel, wherein each of the plurality of buckets could be stored in local storage of said plurality of accelerators; wherein in each iteration, the method comprises: roughly partitioning said data set into a plurality of large buckets; obtaining parameters of said data set that can indicate the distribution of data values in that data set; determining a plurality of data ranges for said data set based on said parameters; and partitioning said plurality of large buckets into a plurality of small buckets corresponding to the plurality of data ranges respectively by using said plurality of accelerators in parallel, wherein each of said plurality of accelerators, for each element in the large bucket it is partitioning, determines a data range to which that element belongs among the plurality of data ranges by computation.