D Computations using GPU

Computations using machines are typically being carried out using some kind of physical processing units. On many systems there are at least two types of processing units that can perform computations: CPUs (central processing units) and GPUs (graphics processing unit), the latter being originally employed for manipulation of images and for game development. While their constituent parts are similar, their architectures differ greatly:

Figuur D.1: The high-level differences between CPUs and GPUs.

In a CPU, the control and the cache are governed centrally in the number of arithmetic logic units (ALUs) is limited. In a GPU, control and cache are more distributed and there are many more ALUs. As a consequence, CPUs are more versatile, ‘smarter’ and faster and work best for singular but complex computations. GPUs, on the other hand, are better equipped to handle massive parallel tasks consisting of repetitive but simple computations.