Current query by humming system can hardly be extended to large massive database as most of them adopt the features extracted from MIDI files which are not widely used and the very time-consuming match methods. In this work, we regard query by humming as a subsequence similarity match problem and exploit the modified SPRING algorithm instead of DTW as the core match method to compare the melody feature extracted from polyphonic music. The SPRING algorithm reduces the algorithm complexity of subsequence match dramatically. Furthermore, we also make it possible for our system to achieve a high speedup over the serial version on different GPU platforms. Experimental results show that our system has more advantages over the state of the art subsequence match method, and has a good scalability to massive database. The processing capacity can reach to thousands of sequences match per second under down sampling. At the same time, the retrieval accuracy results related to all aspects of query by humming point out the encountered problems and the future direction.