Sound Synthesis Using Physical Modeling on Heterogeneous Computing Platforms