Can I reduce the float point precision in Keil to reduce my code size? Thank you.
fixed point is a concept not a type. example instead of 1.00 dollars use 100 cents as an unsigend int 25.0 degres would be 250 as an int. To answer your 1st question. yes you can change the rounding. No that will not save a single byte of RAM or ROM.
Yes, you can save program space and time by using fixed point variables and by using approximations that are just good enough for what you want to achieve. See this PDF: http://www.ganssle.com/approx/approx.pdf And these sites: http://www2.norwich.edu/frey/TaylorPolynomials/ http://www.math.wpi.edu/Course_Materials/Calc3/Labs/node3.html http://math.fullerton.edu/mathews/a2001/Animations/Interpolation/Pade/Pade.html http://www.dspguru.com/comp.dsp/tricks/alg/sincos.htm And you will probably find more if you go searching. The problem with fixed point and approximations is that a good deal more thought and care is required. There is a lot more to go wrong! HOwever, this approach can pay handsome dividends.
"The problem with fixed point and approximations is that a good deal more thought and care is required" Actually, that's a general principle of writing high-performance and/or compact code! Any fool can just jump in and lash-up something, but it takes thought and care to get optimum results! An essential part of that thought and care is, of course, gaining a thorough understanding of the language, the tools, and the processor architecture - none of which is possible without reading the Manuals and the Datasheets!