中文摘要：

数据密集型应用是一类以数据搜索、分析、传输和处理为主的应用。利用GPGPU-SIM模拟器模拟NVIDIA的Fermi架构，分析了Fermi存储结构与数据密集型应用的适应性关系，并给出其存储层次优化设计建议。

英文摘要：

Data-intensive application is a class of application which mainly focuses on searching, analyzing, transporting and processing data. The paper simulates the NVIDIA Fermi architecture using GPGPU-SIM simulator, analyzes the adaptability of Ferrules memory architecture with data-intensive applications,and gives some advices regarding optimization design on memory hierarchy.