I have to deal with real type numbers in my code(synthesisable) so I am using fixed point packages. I have a few doubts in this. 1.I want to define all variables in my function as ufixed(7 downto -6). I believe that all the values i get here dont need more than these number of bits. When i am supposed to find a+b or a/b or a*b , I would type it as resize(a/b , a'high, a'low). Would I get any problems with this format of representation ?? 2. How do i find square root and exponent function of a fixed point number here? I can't use ** in fixed operations as per fixed point package rules.
Guest
#2758876
1. That should work, the result will be rounded/truncated. See http://www.vhdl.org/fphdl/Fixed_ug.pdf. 2. Implement it yourself, or find a core that does it. It's an operation that is too expensive/application-dependent that it would make sense to define it as a VHDL operator.
Okay.Thanks for replying :)
How do i find b^a when i want to write as a synthesisable code??? Please give me an Idea.