โ Computer Organization & Architecture learning pathComputer Organization & Architecture ยท Module 2 of 10
Data representation and arithmetic
Integers, floating point and ALU design.
๐ Key ideas
- Signed integers are stored in 2's complement; overflow happens when the result does not fit.
- Booth's algorithm multiplies signed numbers efficiently.
- Floating point (IEEE 754) stores sign, exponent and mantissa; single precision is 32 bits, double is 64.
- Carry look-ahead adders compute carries in parallel, faster than ripple-carry.
- Floating-point numbers cannot represent all decimals exactly, so 0.1 + 0.2 is not exactly 0.3.
๐ก Real-life example: Floating point is like scientific notation (6.02 x 10^23): a few digits of precision plus a power that sets the size.
IEEE 754Standard for floating-point numbers
MantissaSignificant digits of a floating-point number
OverflowResult too large for the bit width
Booth's algorithmMethod for signed binary multiplication