The basic verification Python script contained in this folder works as follows:
Reads the input data from the text files named
PLIO_i[*]in thev_inputfolder.This data follows the AIE-ML Kernel and Graph programming User Guide (UG1603)
The
v_inputfolder has two other folders: one contains text files representing a sampled rectangular wave, the other contains text files with random data. You can replace the text files in thev_inputfolder with those.
Reads the AI Engine simulation output data from the text files named
PLIO_o[*]in thev_outputfolder.This data has been copied and pasted from the simulation outputs created with the AMD Vitis™ Unified IDE.
Performs the FFT on the input data with the NumPy library functions to create a golden reference to compare with the AI Engine simulation outputs.
Creates a graph that plots instance 0 for both the AI Engine output data and the NumPy golden reference.
Performs the iFFT on the AI Engine simulation output data with the NumPy library functions to compare it with the input data.
Creates a graph that plots instance 0 for both the NumPy iFFT of the AI Engine outputs and the input data.
Creates a graph displaying the relative error between the NumPy golden reference FFT and the AI Engine one for each point of instance 0.
Inside the script there is a section where it is possible to set the following parameters:
The datatype
The number of points
The number of signal parallel instances
The I/O temporal interleaving factor \(\theta\) which is the ratio between the PL clock speed and the sample rate
The I/O spacial interleaving factor which is the amount of samples packed together in a I/O channel
The number of sampling repetitions: the parameter inside the AI Engine Run method in the graph .cpp file
Do not change such settings parameters while reproducing this tutorial.