When writing and using AI software it is typical to use a whole software stack to build a solution. To help reduce the risk of incompatibility NVIDIA AI Enterprise Software Suite can be thought of as a container mounted on a virtualized platform. It contains all the environmental variables and dependencies you need to run a solution.
The AI and data science tools and frameworks are on top of the stack and include programs. These include TensorFlow, PyTorch, NVIDIA; TensorRT, Triton Interface Server, and RAPIDS.
For cloud and remote deployments, NVIDIA GPU Operator and Network Operator are also provided. Useful for when mobile applications leverage heavy metal to process data before returning it in real-time to the lightweight application.
The last part of NVIDIA's three-part offering is an infrastructure optimization set of tools. These are to help ensure a constant and performant uptime is achieved throughout use. The three core programs are NVIDIA vGPU, Magnum IO, and CUDA-X AI.