Method and apparatus for converting from integer to floating point representation
Abstract:
Apparatus and methods for conversion from signed integer to a floating point representation are provided. Two's complementation and lead zero count operations are performed in parallel. Exponent generation and mantissa shifting are performed in parallel. Generation of the floating point exponent from the signed integer, including application of a scaling factor, is performed using a 3:2 compressor or carry-save adder and an adder. Two's complementation for generation of the mantissa in unsigned integer format is performed using an adder. Lead zero count for controlling mantissa shifting is performed by one's complementing the signed integer if negative, counting lead zeros in the one's complement output, and determining, using the one's complement output, whether the one's complement lead zero count differs from the two's complement lead zero count by one.
Information query
Patent Agency Ranking
0/0