Floating Point & Fixed Point Microprocessor

[popat_panjaru]

hi to all

can u explain what is the difference between floating point and fixed point microprocessr?..same for microcontroller if it is..........


Popatlal

 

[foxabilo]

Fixed point maths is simply 12345.123 where the decimal place remains in the same "virtual" location at all times, all mathematics done in this fashion can be achived with simple hardware, often called hardware multiply. Floating point numbers have a fixed data size i.e 6 bytes but the decimal place can be moved so if you have a small number you want very accuraly you can move the decimal place i.e 0.12743637 or you can store very large numbers with little accuracy 2638364.3, the other addition is that of exponents 1.6e+09 which allows very very small/large numbers to be encoded and used, the downside is that floating point manipulation is much more complex and requires either more software (slower) or more hardware (expensive)

You will find that all hardware floating point capable micro controllers cost more that fixed point ones, although you can perform software floating point in any micro controller with software.

Most uses of floating point can be removed with proper analyse of the system being controlled, many implementations of floating point mathematics are unnecessary if you have a good idea of the size and range of numbers you expect to get in your end application. For example, I have seen people use a floating point system when measuring and converting Fahrenheit to Centigrade when a simple manipulation of the numbers allows fixed point or lookup tables to be used over the small range of temperatures their unit measures.

In simple terms, floating point, if mentioned specifically in the micro controller documentation usually refers to hardware (quick) floating point and fixed point just means a standard hardware multiply ala PIC18.

Hope that helps