6.6.3.1.2 NaN Operands with Float-to-Integer Conversion

The FF2DI and FF2LI are the float-to-integer instructions. These instructions can output Huge Integer in lieu of Invalid when the source is a value that would convert to an integer outside the range of the result format under the applicable rounding attribute. This output is implemented as a new exception, Huge Integer (FSR.HUGI).

The IEEE 754-2008/2019 standard calls for Invalid to be signaled if this situation cannot otherwise be indicated. The FSR.HUGI exception is considered an implementation of “otherwise indicated,” making the FF2DI and FF2LI instructions compliant.

The module drives status output Invalid (and does not drive Huge Integer) when the source is ±NaN, or ±∞ per the IEEE 754-2008/2019 standards. Note that Invalid is also driven for a qNaN input