Data-intensive applications have become one of the most important computer applications today due to increasing requirements of user demand. To get greater application performance it needs greater computing capability. The solution of this lies in MapReduce parallel programming model with many-core micro architecture. MapReduce was created by google for application development on data centers with thousands of servers. But the problem with present many-core micro architecture is that it does not match MapReduce runtime which makes many-core chip system unable to get deserving computing capability. In order to achieve huge computing and processing capability in high performance many-core chip system, this research combines many-core and cloud computing, two latest directions in computer science, in terms of computer micro architecture and studies how to design many-core 3D memory architecture to meet the demand of memory performance in micro architecture for MapReduce.