Energy consumption is a crucial factor that constrains applications running on different computer architectures and platforms. Video encoding is no exception and it is relevant in mobile platforms. In the video encoders memory system the energy consumption is mainly due to the increase in bandwidth and the storage of the reference frames used in the encoding process of the video. The motion estimation stage is responsible for this process, which requires large amount of access to the internal and external memories of the chip, which leads to high energy consumption for video encoding. This paper presents an evaluation of different schemes with data reuse using different memory technologies in the memory hierarchy, including nonvolatile memories (NVM), such as: PCRAM, ReRAM and STT-RAM. Our results show that schemes composed by DRAM and NVM present energy savings of 43%, on average, when compared to conventional DRAM and SRAM schemes. Furthermore, the best results were obtained with ReRAM as external memory and ReRAM or STT-RAM as the scratch-pad memory (SPM), such configurations reach a total memory energy consumption reduction of 89% and 87%, respectively, when compared to the conventional configuration of external DRAM and on-chip SRAM memories.