It is also super fast on architectures with a FPU! The code for the encoder has been written a long time ago (see below) and the only work done on this fork consists of reorganizing the code and ...