The question is based on PIC Architecture in microprocessors.

Floating point representations start by reducing the number to scientific notation in binary. Then the sign, the exponent, and the mantissa of the number are independently ``code"d . The method used for coding each part of the representation can vary.

One 8-bit floating point representation from http://www.cburch.com/books/float/index.html#s2 is specified as (1,4,3,-7): 

• [bit-7] represents the sign (1 for negative, 0 for positive), 
• [bit-6:bit3] represents the binary exponent (the sum of 7 and the actual exponent),
• [bit-2:bit0] represents the mantissa (the mantissa's fractional part).

1. Identify the maximum and minimum numbers that can be represented.
2. Which number is represented by 8-bit value 110101012110101012? Show your reasoning.
3. What is the 8-bit value that represents 10.821010.8210? Show your reasoning.
4. Plan a routine (using HIPO) labelled gtman that will compare two 8-bit floating point numbers. You do not need to write code. 
5. 
   
   • The input 8-bit values are stored in 8-bit registers labelled RegA and RegB.
   • Carry flag is set if 8-bit value in RegA has a larger mantissa than RegB; else Carry flag is clear