Methods and apparatus for determining a floating-point exponent associated with an underflow condition or an overflow condition

An apparatus structured to determine a floating-point exponent, the apparatus comprising:

• a floating-point hardware unit;
• a floating point software unit;
• and
• an exponent determination module, operatively coupled to the floating-point hardware unit and the floating-point software unit, the exponent determination module being structured to:
• receive a truncated exponent from the floating-point hardware unit;
• add an underflow shifting constant to the truncated exponent to produce a first result if an underflow condition occurred;
• truncate the first result to a predetermined number of least significant bits to produce a second result if the underflow condition occurred;
• subtract the underflow shifting constant from the second result to produce the floating-point exponent if the underflow condition occurred;
• and
• pass the floating-point exponent to the floating point software unit.

A machine accessible medium storing instructions, the instructions being structured to cause an apparatus to:

• receive a truncated exponent from a floating-point hardware unit;
• subtract an overflow shifting constant from the truncated exponent to produce a first result if an overflow condition occurred;
• truncate the first result to a predetermined number of least significant bits to produce a second result if the overflow condition occurred;
• and
• add the overflow shifting constant to the second result if the overflow condition occurred.

A method of determining a floating-point exponent associated with an overflow condition, the method comprising:

• receiving a biased exponent from a hardware floating-point unit;
• subtracting a shifting constant from the biased exponent to produce a first result,
• truncating the first result to a predetermined number of least significant bits to produce a second result;
• and
• adding the shifting constant to the second result to produce the floating-point exponent.

A method of determining a floating-point exponent associated with an underflow condition, the method comprising:

• receiving a biased exponent from a hardware floating-point unit;
• adding a shifting constant to the biased exponent to produce a first result,
• truncating the first result to a predetermined number of least significant bits to produce a second result;
• and
• subtracting the shifting constant from the second result to produce the floating-point exponent.