Coordinated multi-point transmission/reception (CoMP) has been proposed as an effective technology to improve cell-edge throughput in next-generation wireless systems. However, all the BSs(Base Station) need to share all the users' messages to do the global precoding for supporting each user best service, such as which BS will be chosen for cooperating or how to realize the cooperation, lots of energy will be consumed when doing this process. In addition, in most of the existing works, BSs always allocate equal power to all the users neglecting user's Channel State Informations(CSIs) or their service requirements, this may also cause enery wasting. In this paper, we focus on the throughput maximizing problem while fully considering energy saving problem in the CoMP systems. At first, we propose the concept of “cell zooming” which we define that one of the cooperating BSs will “sleep” if it improves little to user's service, then a sub-carrier allocation algorithm with varying power assignment in downlink CoMP system is put forward, this can lower the transmit power of BS. We formulate that the sub-carrier and power allocation problem is a mixed binary integer programming (MBIP) problem, a “green” subcarrier and power allocation algorithm which can maximize the cell's throughput is proposed at last, the power allocated to each user will be calculated. Simulation results show that the proposed sub-carrier and power allocation algorithm is much better on users' throughput and fairness while consumes less energy than the existing ones.