need to subtract two numbers/signals, those have different data types -floating-point and fixed-point data types-

[kikibillel52]

Hello,

i need to subtracte two signal, those have different data types: "first_signal: floating-point data type" ;   "second_signal: fixed-point data type".

i am going to prepare a vhdl code to calculate the subtraction of these two inputs. i would like to hear from you and get some suggestion, like what i need to avoid or to take into consideration.

I am very glad to hear from you and thanks in advance.

[drjohnsmith]

How are u handling floating point .

[kikibillel52]

good morning drjohsmith

thank you very much for your comment.

so, i have two system doing the same function (multiplication), one is prepared in vhdl code and using no floating-point library (just fixed-point data type). The second system is implemented in cyclone V with ARM proccessor, and the floating point library is available (i will use the single precision - 32 bits). 32 bits precision equivalent to 7 decimal digits.

So, the both systems doing the same operation, have the same results with different data type.

Do you have any suggestions that i can follow to do this kind of comparison and thanks in advance.

LG, Billel 

[drjohnsmith]

can I ask how your using a cyclone V on a Xilinx Part ?

[kikibillel52]

that what is available now, my Problem is how do this comparison between two different data types, not how to use Cyclone V, that is not my fucus, it can be also XILINX, and getting the same output data type (Floating point).
Thanks in advance.

[drjohnsmith]

Just a reminder this is a forum run by Xilinx for Xilxin support,

[kikibillel52]

thanks for your advice.

if you have any suggestions regarding the issue, i would like to hear from you.

[u4223374]

Buy a Xilinx chip? That would make it easy. Xilinx has their Floating Point IP core, and Xilinx's Vivado HLS can easily accept both fixed-point and floating-point values, do maths on them, and spit out a result.

At the highest level, the easiest approach here is probably to get the fixed-point values into the PS, convert them to floating-point (easy), and then do the comparison in C code. It'll be easier to get the PS to handle fixed-point than getting the PL to handle floating-point. By far the hardest part is likely to be getting the values into the PS - and you won't get much help for that here, since it'll be manufacturer-specific and everyone here just knows how to do it with Xilinx chips.

[kikibillel52]

thanks a lot for your suggestions. a very helpful.