Getting started with GPUs

To begin, you'll need a computer with a discrete GPU installed. For the purposes of this chapter, that will mean a GPU manufactured by NVIDIA. While other brands of GPUs are available, support for those is much weaker, both for general scientific programs, and, in particular, in Julia. Hence, we will not have the opportunity to discuss them in this chapter and we limit our focus to NVIDIA GPUs. 

Once you have access to a computer with a GPU (whether on your desktop or on the cloud), the first order of business is to install a driver and CUDA toolkit on the machine. The exact instructions will depend on your operating system and the GPU model you use. You will need to ensure that the versions of the device driver and CUDA toolkit are compatible.

CUDA is the programming model used by NVIDIA for general-purpose computing on its GPUs. It&apos...