Is increased precision of intermediate results of floating-point computations legal?

Printable View