We have developed a new way of computing three-point functions. This is based on a factorization with open spin- (and to some extent flavour-) indices that allows us to generate these simultaneously for many different initial and final states and momenta, with little computational overhead.

This factorization as well as its implementation will be explained. A new library has been written to efficiently provide the necessary functionality on modern parallel architectures and on CPUs, including Intel's Xeon Phi series. Key ideas and the user interface will be explained.