Using a persistent variable array in MATLAB function for HDL Coder

I have MATLAB function which calculates the two maximum numbers in an array of 4096 elements. I have a persistent array - of int32 numbers - which receives a single element from the function block and assigns it to the corresponding index. However when I try to build HDL for the subsystem that contains the MATLAB function. I receive the error "The persistent variable 'array_per' in chart '....' is of non-integer type. The option 'Support floating-point numbers' has not been set Code Genaration configuration options.

Is there a way to fix that?

Answer 1

[Tom Richter]

Hi Kaan,

it would be good if you could share your MATLAB code you try to implement using the MATLAB Function block. For me it sounds like a wrong data type setting. The array should be initialized like this:

persistent array_per

if isempty(array_per)
    array_per = zeros(4096,1,'int32');
end

The array will keep the data type as long as you index only one element:

array_per(4) = input_val;

Such a big array is better implemented using RAM (only one read/write action per function call).

I also doubt that you need such a big persistent variable to get the two most max values. I would use:

• An index counter (persistent variable)
• two persistent variables for the index (index_max1, index_max2)
• two persistent variables for the max values (value_max1, value_max2)
• You compare the input_val with value_max2 and value_max1
• You assign value_max1 to value_max2 and index_max1 to index_max2 if input_val > value_max1 and then the input_val to value_max1 and index_count to index_max1
• ...
• ...
• You output results and reset all persistent variables to 0 if counter reaches max index value (e.g. 4096)

The rest is simple I guess.

I hope that helps,

Tom