Introduction

The mapped_allocator allocator defined in mapped_allocator.hpp can be used to allocate host mapped memory from the NVIDIA CUDA API. Behind the scenes, the mapped_allocator uses cudaHostAlloc and cudaFreeHost.

The demonstration code is based on example from Dr. Dobb.