Main Content

Convert Code Containing Global Variables to Fixed-Point

This example shows how to convert a MATLAB® algorithm containing global variables to fixed point using the Fixed-Point Converter app.

  1. In a local writable folder, create the function use_globals.m.

    function y = use_globals(u)
    %#codegen
    % Declare AR and B as global variables
    global AR;
    global B;
    AR(1) = u + B(1);
    y = AR * 2;

  2. In the same folder, create a test file, use_globals_tb.m that calls the function.

    u = 55;
    global AR B;
    AR = ones(4);
    B=[1 2 3];
    y = use_globals(u);

  3. On the MATLAB toolstrip, in the Apps tab, under Code Generation, click the Fixed-Point Converter app icon.

  4. To add the entry-point function, use_globals.m to the project, on the Select Source Files page, browse to the file, and click Open. Click Next.

  5. On the Define Input Types page, add use_globals_tb.m as the test file. Click Autodefine Input Types.

    The app determines from the test file that the input type of the input u is double(1x1).

  6. Select Yes next to Does this code use global variables. By default, the Fixed-Point Converter app names the first global variable in a project g.

  7. Type in the names of the global variables in your code. In the field to the right of the global variable AR, specify its type as double(4x4).

  8. The global variable B is not assigned in the use_globals function. Define this variable as a global constant by clicking the field to the right of the constant and selecting Define Constant Value. Type in the value of B as it is defined in the test file, [1 2 3]. The app indicates that B has the value[1 2 3]. The app indicates that AR is not initialized.

  9. Click Next. The app generates an instrumented MEX function for your entry-point MATLAB function. On the Convert to Fixed-Point page, click Simulate to simulate the function, gather range information, and get proposed data types.

  10. Click Convert to accept the proposed data types and convert the function to fixed-point.

    In the generated fixed-point code, the global variable AR is now AR_g.

    The wrapper function contains three global variables: AR, AR_g, and B, where AR_g is set equal to a fi-casted AR, and AR is set equal to a double casted AR_g. The global variable B does not have a separate variable in the fixed-point code because it is a constant.

    function y = use_globals_fixpt(u)
    %#codegen
    % Declare AR and B as global variables
    fm = get_fimath();
    
    global AR_g;
    global B;
    AR_g(1) = fi(u + B(1), 0, 6, 0, fm);
    y = fi(AR_g * fi(2, 0, 2, 0, fm), 0, 7, 0, fm);
    end
    
    
    function fm = get_fimath()
    	fm = fimath('RoundingMethod', 'Floor',...
     'OverflowAction', 'Wrap', 'ProductMode', 'FullPrecision',...
     'MaxProductWordLength', 128, 'SumMode', 'FullPrecision',...
     'MaxSumWordLength', 128);
    end
    

Related Topics