Fixed point assigns a permanent position to the binary point. Floating point stores a sign, exponent and significand so the point can effectively move. Fixed point offers predictable scaling; floating point offers a much wider range but still has finite precision.
Worked example
An unsigned eight-bit fixed-point value with four fractional bits stores 00110100 as 52 divided by 16 = 3.25. Its step size is 1/16. The same real value can be written as binary 1.101 × 2^1, illustrating the normalized form used by floating point.
What to watch for
A bit pattern alone does not specify a fixed-point scale. Floating point also includes special values and rounding rules; a simple normalized example is not a complete encoding specification.
Try it yourself
With three fractional bits, what stored integer represents 2.5?
Answer: Multiply by 2^3: 2.5 × 8 = 20. The value is exactly representable because 2.5 is a multiple of 1/8.
How to study this topic
Write the base and bit width beside every value. Expand place weights before using a shortcut, then convert your answer back to decimal as an independent check. Representation and value are different ideas.
Explore related programmes
These recorded programmes are upcoming. This article is general educational content, not a syllabus or course sample.
Digital Electronics