This data type is a floating point value that uses the natural size of the host architecture to store its value.  On 32-bit systems, this uses float for storage and on 64-bit systems it uses double for storage.  

As an optimization, the compiler and runtime have special knowledge of this data type, so while there is no dedicated set of IL instructions to deal with variable-size floating point values, the runtime will transform uses of nfloat into their native underlying representation transparently.

This type is defined in lowercase because it makes it feel like a C# data type when used in C# source code.