Starting in R2022a, HDL Coder supports matrix types inputs and outputs at the DUT interface. This enhancement reduces the overhead of adding Reshape and Concatenate blocks inside the DUT and also improves readability of the generated HDL code.
I assume you talk about a Simulink model and with HDL model you mean a subsystem (DUT) which contains the algorithm for HDL code generation. With that you can e.g. use a constant block which outputs a matrix and use it as an input to the DUT. If the DUT is a model with imports and outposts you can define to port size (e.g [3, 2]) and the data type (e.g. uint8 or fixdt(1,8,4)). The data type is the same for all matrix entries.
Where did you try to include the [uint8, uint8] term?
