Floating point unit equipped also to perform integer addition as well as floating point to integer conversion

A machine-readable storage medium having stored thereon formal descriptions describing a floating point unit to perform floating point operations and integer operations, wherein when synthesized said formal descriptions yield a gate-level description of an integrated circuit including:

• an alignment unit, to receive a first input and a second input, and to align the first and second inputs;
• a plurality of computational units, coupled to the alignment unit, to receive and add the aligned first and second inputs;
• a first bypass datapath to bypass a predetermined one of the aligned first and second inputs around the plurality of computational units;
• and
• a first bypass multiplexer, coupled with the plurality of computational units and the first bypass datapath, to select the first bypass datapath for a floating point to integer conversion if the other one of the aligned first and second inputs is an integer conversion factor, and to select the plurality of computational units for a floating point addition if the other one of the aligned first and second inputs is a floating point input.

An apparatus comprising:

• a first data path comprising a shifter and an adder to add two floating point numbers;
• a second data path comprising the shifter and bypassing the adder to convert a floating point number to an integer;
• and
• a third data path comprising the adder and bypassing the shifter to add two integers.

A machine-readable storage medium having stored thereon formal descriptions that when synthesized yield a lower-level description of an integrated circuit comprising:

• a first data path comprising a shifter and an adder to add two floating point numbers;
• a second data path comprising the shifter and bypassing the adder to convert a floating point number to an integer;
• and
• a third data path comprising the adder and bypassing the shifter to add two integers.

A method comprising:

• aligning and adding a first input and a second input if the first input and the second input are floating point numbers;
• aligning the first input and the second input and bypassing addition of the first input and the second input if the first input is a conversion factor and the second input is a floating point number;
• and
• adding the first input and the second input and bypassing alignment of the first input and the second input if the first input and the second input are integers.