This is the patent at issue:
> Method and apparatus for handling overflow and underflow in processing floating-point numbers
> Method and apparatus for handling overflow and underflow in processing floating-point numbers
> In an effort to unify methods employed in computer systems for performing binary floating-point arithmetic, the IEEE in the early 1980's standardized computer floating-point numbers. > [...] > In reshaping the IEEE format, the present invention retains most of its benefits while offering the microprocessor designer and the real-time programmer a much cleaner and more efficient implementation.
For what I understand, is an extension of the IEEE standard.
But this is still perplexing: Surely, Linux just uses IEEE754 floating point, which predates this patent by 15 years?