R Tree Window Query Algorithm Based On Multi-core Platform Is Optimized

With the development of the national economy information technology, spatial information processing becomes more and more important in the human socio-economic development, national security, and many other fields. Digital City, Digital Valley, Digital Enterprise, Digital Cultural Heritage, Mobile Location-based Services, Virtual Reality and visualization of three-dimensional dynamic applications and so on are heavily dependent on the fast accessing and processing of massive spatial information. The accessing and processing of massive spatial information are not only the data-intensive operations but also the CPU-intensive operations. With Moore’s Law begin to lose its effect on the CPU frequency, a dual-core CPU appears, which integrates with two CPUs on a chip, and will continue to integrate more CPUs. In multi-core platform, the thread will be really parallel executed, rather than executed in turn on single-core platform, the thread’s load balancing(also CPU’s load balancing) and the cache’s efficient utilization become the important key to improve the multi-core platform’s performance. However, traditional serial algorithm can not meet this requirement properly. The parallel algorithm study based on multi-core becomes a new hotpot.This article is researched under this background, which analysis and summarizes the hardware and software technology based on the multi-core platform, and includes the principle of the task queue model. and the most notable change is the OpenMP3.0API mainly used in the sharing parallel architecture, which introduced the task queue model. And it is claimed to resolve the parallelization of the complex data structure and recursive structure. Mean while the paper try to used the task queue model in the classic spatial index algorithm-R-tree Index for the study, and develop an integrated model for RTree window query algorithm. And use different types and scales of data sets to test the improved algorithm on a dual-core platform. At last, we get the algorithm speed up ratio on the dual-core platform, and came to a conclusion that the performance of the improved algorithm can be generally increased by20%.