Testing Against the Python Model

Once you have the module passing a single test, we can run the test the module against the test vectors generated from the python golden module. We will use the testbench in tb_cubic.sv. This file will read the test vector inputs, instantiate your module as a DUT, run the inputs through your DUT module and compare the DUT outputs with the outputs from the python golden model.

At the beginning of the testbench file you will see:

module tb_cubic;

    // Parameters
    localparam int WID = 16;
    localparam int FBITS = 8;
    ...

Change FBITS=8 or FBITS=12 as needed. I would first test FBITS=8 since this test case is easier to match. Once you have selected the correct parameters, the testbench can be run by using the xilinxutils function:

Open a terminal in Unix or command window in Windows (On Windows, you cannot use Powershell)
Activate the virtual environment with the xilinxutils package
Follow the instructions to set the path for Vivado tools
Navigate to the hwdesign/labs/subc directory
Run

sv_sim --source cubic.sv --tb tb_cubic.sv

This will run the three steps in synthesizing and simulating the SV mdule. The outputs will be stored in a CSV file, test_outputs/tv_w16_f8_sv.csv or test_outputs/tv_w16_f12_sv.csv depending on the choice of FBITS. You should see how many tests passed, and you can keep modifying the SV code until all test passed.

You can also follow the specialized python instructions for running the the tests on the NYU server.

Verify the test outputs

Once you have the tests passing, you can run the program:

python run_tests --tests sv_f8 
python run_tests --tests sv_f12 

These commands will tests that the test vectors match.

If you are running Vitis and Vivado on the NYU machine, you will need to follow the instructions for setting up uv. Then, activate the hwdesign virtual environment and run:

uv run python run_tests --tests sv_f8 
un run python run_tests --tests sv_f12 

Go to submission