Consider numbers stored in a floating point format with the exponent encoded as a 3-bit unsigned integer equal to the actual exponent plus three. The normalized mantissa uses signed magnitude encoding with 1 bit for the sign, 4 bits for the fraction, and an implicit 1" left of the binary point. Two numbers, x and y are stored in the above format as 00100011 and 11100000, respectively. Find the 8-bit binary word for the sum s = x + y, where s conforms to the above format. Solve for s by decoding x and y to decimal, do the sum, and then encode the sum using rounding and conforming to the above format. Determine the absolute error of the stored sum.