Modular Physical Modeling Synthesis Environments on GPU

Physical mode ling synthesis is a powerful means of access to a wide variety of synthetic sounds of an acoustic character—one longstanding design principle underlying such methods has been, and continues to be modularity, or the decomposition of a complex instrument into simpler building blocks. In this paper, various modular physical modeling design environments, based on the use of time stepping methods such as finite difference time domain methods are described, with an emphasis on the underlying computational behaviour of such methods, both in the run-time loop and in precomputation. As such methods are computationally intensive, additional emphasis is placed on issues surrounding parallelisation, and implementation in highly parallel hardware such as graphics processing units. This paper is paired with a recently completed multichannel piece, and the composer’s perspective on working with such environments is also addressed.