NVIDIA CUDA Tutorial 5: Memory Overview

The GPU has a complicated but flexible set of different memories. It's sometimes called a memory hierarchy. Each type of memory has various characteristics which we'll look into individually later. This tutorial is a short introduction to all the memories. These memories can be confusing. In many instances, they are the same stuff, for instance, the L1 cache and shared memory are the same bytes. The texture, global, constant and local memory are the same bytes as well. At the end I've put a small demonstration of the NVidia Visual Profiler, a program which I think everybody that codes CUDA should definitely have a look at. The website with the Device Query program is: http://docs.nvidia.com/cuda/cuda-samples/