I have decimal numbers ( lets say 8.432, 5.256 etc) in my MATLAB as input for arithmetic operations ,I convert my code to vhdl using HDL Coder. The data type i chose for input and output blocks in Simulink is int32. The code is converted to vhdl but when I see the binary output I only have the non decimal part i.e 8 and 5 in above example .How I could get the binary output for the decimal part too.May be its simple question but being very new in this field dnt know how to do this.
You can try to use the data type fixdt if this type is available. Two years ago i generate some HDL code with the HDL-Coder and i use this type.For example, with fixdt(0,16,0) you can create a register which represents a 16bit In- or Output. Maybe this would help you with your problem.
The data type you have chosen for input and output blocks, int32 can only represent (signed) integer values, so any non-integer inputs or outputs will be converted to integers with no fractional part.
Does your VHDL code use fixed or floating point variables for calculations? If not, then you will need to find some other means to handle the fractional part.
Note that by default Matlab stores numbers using the IEEE Standard 754 double precision representation, which has a binary (base 2) mantissa and binary exponent. If you need an exact decimal (base 10) calculation, there may be small rounding errors. You could get round this using BCD (binary coded decimal) arithmetic.
I don't use HDL coder or VHDL, so I can't help with specific details.
http://www.eda.org/fphdl/Float_ug.pdf
http://www.eda-stds.org/fphdl/Fixed_ug.pdf
http://www.eda.org/fphdl/fpfaq.html
- Badges
- Science topic