As the quality of life continues to increase, so does the demand for high-quality multi-views. For this reason, the Multiview HEVC (MV-HEVC) extension was completed in July 2014. Considering the high complexity of the MV-HEVC coding algorithm and the SoC without the mobile end temporarily using the IP of the MV-HEVC decompressor, it is very effective to complete the MV-HEVC real-time decompression task above 1088p by using only CPU. Hence, we design a novel heterogeneous cluster of the universal mobile processor platform. Because the most difficult task in a heterogeneous cluster system is to schedule tasks, it keeps the entire cluster system in balance, but often needs to combine the data types of scheduled tasks. Therefore, a support vector machine prediction method based on new Wavefront Parallel Processing method for MV-HEVC and mobile platform features is proposed to improve the task allocation accuracy of a MV-HEVC heterogeneous cluster system.