Accelerate Computing with CUDA Python

NVIDIA is unifying the CUDA Python ecosystem with a single standard low-level API that provides access to the CUDA host APIs directly from Python. Recently released on GitHub, CUDA Python provides a foundation for the ecosystem to build libraries that target NVIDIA GPUs. We'll discuss the challenges that motivated the project and give an overview of the library features, illustrated by an example workflow. We'll also discuss the experience and benefits that developers have had by adding support for CUDA Python to their projects.