发明名称 System, Method, and Computer Program Product for Efficient Aggregation of Data Records of Big Data
摘要 A computerized method, system, and computer program product can allow for processing by a processor data records (xDRs). The method can include receiving by the processor xDRs, each xDR including a vector of unique data fields, the xDRs forming together a stream. The stream can be stored by the processor in a memory of a general purpose graphic processor unit (GPGPU). An index vector of xDRs can be created by the processor, wherein each xDR is assigned a unique index identifier, and stored by the processor in a memory of the GPGPU. The index vector of xDRs is sorted with respect to at least a key of at least a unique data element selected. The sorted index vector can be stored in the memory of the GPGPU. A selection of at least the unique data field can be performed and aggregation of the stream performed with respect to the selection.
申请公布号 US2015278269(A1) 申请公布日期 2015.10.01
申请号 US201514672230 申请日期 2015.03.30
申请人 TEOCO CORPORATION 发明人 SAPIR Yoav;CAHANA Ori
分类号 G06F17/30;G06T1/20 主分类号 G06F17/30
代理机构 代理人
主权项 1. A computerized method for processing, by at least one processor, a plurality of data records (DRs), the method comprising: receiving, by the at least one processor, a plurality of DRs, each DR comprising a vector of unique data fields; storing, by the at least one processor, the plurality of DRs in a primary memory; creating, by the at least one processor, an index vector of DRs, wherein each DR of the plurality of DRs is assigned a unique index identifier; storing the index vector of DRs in a memory of a general purpose graphic processing unit (GPGPU); sorting the index vector of DRs by the GPGPU related to at least a key of the at least a unique data field; and storing the sorted index vector of DRs in the memory of the GPGPU.
地址 Fairfax VA US