Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
What's the biggest difference among fixed point and floating point arithmetic? Where can i find introductory tutorials about floating point arithmetic?
With same number of bits, floating point give a better accuracy compare with fixed point. For example 0.000000000345678 need more bits to store than 0.345678e-9.
Sorry but previous reply is not true. Fixed point is more accurate and faster but has a limited range of values. Floating point is slower and loss of accuracy (for the same number of bits) but has a large range of values. example 32bit fixed point can represent 4 billion values accurate to the number of bits. 32 bit float will give 1 bit sign 8 bit power and 24bit mantisa (from memory) which will allow 1e128 to - 1e128 (also down to fractional values 1e-127) but the 24 bit mantisa gives an accuracy of 1 in 16.7 million. Also it is hard to figure out inaccuracies in the float calculations but they will be consistant. Where there are no inaccuracies in the floating point outside of overflow and loss of precission and they are easily figured out.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
