Compiling fixedpt converted code into VHDL

We have a matlab code that we compile into VHDL. We have a make file that did this programmatically using something like

codegen -float2fixed fixptcfg -config hdlcfg -args {...} function_name

Since we need more control over fixedpt conversion, we now first run the script made by

fixedPointConverter -tocode spectrometer -script fixpt_conversion

and then we directly compile the generated code with something like:

addpath('./codegen/spectrometer/fixpt')
codegen -config hdlcfg -args {int16(0),int16(0)} spectrometer_fixpt

I have three questions:

1. Is this the right way to do it?
2. How does it deal with test_bench? The supplied test_bench of course calls spectrometer, not spectrometer_fixpt. Would it automatically adapt it? Or do I need to manually create a version of test bench that calls spectrometer_fixpt?
3. How can I compile a subroutine inside spectrometer_fixpt? spectrometer calls function, I want to compile them individually, without having to rerun fixpt converson manually for each of them. Is that possible?

Answer 1

[Kiran Kintali]

https://www.mathworks.com/help/hdlcoder/gs/generate-hdl-code-from-matlab-code-using-the-command-line-interface.html

Generate HDL Code from MATLAB Code Using the Command Line Interface

This example shows how to use the HDL Coder™ command line interface to generate HDL code from MATLAB® code, including floating-point to fixed-point conversion and FPGA programming file generation.

[Anze Slosar]

Following that example is how we did it earlier. But that seemed to have very little options to finetune fixed point conversion.

But now I see that configuation options for

fiaccel -float2fixed

are exactly the same object as for codegen, so I can just use addTypeSpecification to set some fixed points manually. Which is what I was kinda looking for, thanks.