VHDL code for complex arithmetics

[vijayalaxmi]

i want to know a synthesizable code to find the value of....
for eg. x to the power of y , where x is an integer and y is a real number ,like
2 to the power of 0.5.

i tried with ** operator,like defining x and y as integer and real data types respectively and z as areal data type .given the exp. z<="**"(x,y),
where "**" is a predefined function in math_real package in ieee library.
but real data types are not synthesizable.
so,plz someone help me in this regard.

 

[aajizattari]

vhdl code

i TOO am a beginner here but

I think one possible way to step forward is


looking into FIXED POINT ARITHMETIC


May be you can get approximate solutions

 

[echo47]

vhdl code

Also check the libraries and cores provided with your synthesis software. It may contain floating point math modules that you can drop into your project.

 

[avimit]

Re: vhdl code

Dear vijayalaxmi,
What you are trying to do isn't possible. This is because the synthesis flow is not mature enough to cater for this kind of complex arithmatic.
So, you need to see what 'y' is, then build a customised block for finding x**y, may be you will have to use a rom if range of x is limited, and store the values of x**y in rom.
For example in mp3 decoding you need to find z=x**(4/3) so I had do to a customized block to implement that, where I store some values in ROM, and the rest was a linear interpolation.
Hope it helps
Kr,
Avi
http://www.vlsiip.com

 

[suru]

vhdl code

i need vhdl code of turbo code. plz help me to get it.

 

[vinodkumar]

Re: vhdl code

hi,is this operation required every time or its required only once.and also the power is multiple of 2 or anything else.