Test Vector Comparison - 2023.2 English

Vitis Tutorials: AI Engine (XD100)

Document ID
Release Date
2023.2 English

When comparing simulation results against test vectors, a MATLAB script is invoked to perform the processing. An example of a successful comparison is shown in Figure 11.


Figure 11 - Simulation Verification

The output provides three different indications of simulation performance. The first is an indication of whether the simulation output matched the corresponding test vector. There will be one comparison for each kernel simulated. The script compares int32 values which represent either output pixel indexes or floating-point interpolated pixel values. Since there will be slight variations in floating point calculations, the comparison allows for a difference in the two least significant mantissa bits of the floating-point number.

The second comparison indicates maximum pixel value difference between AIE simulation results and single precision MATLAB generated vectors. Pixels take on values in the range [0, 255], and this result provides the maximum of the differences between all pairs of corresponding pixels.

The final comparison indicates the maximum pixel value difference between AI Engine simulation results and double precision floating-point results generated by the MATLAB interp2 function.