Software engineering is a complex process. Common expectations of scientific software include functionality, flexibility and robustness. At the early stages of developing a scientific approach, these aims should take priority, and premature optimization for speed or other hardware resources tends to be distracting. Working in a high-level language such as R is therefore a rapid and effective choice. There is plenty of time, once the right approach has been settled on, to worry about whether the computation really needs to be faster and where the bottleneck lies.
