PSS computes the continuous time spectrum of the output voltage. This differs from a DFT or FFT, which is probably what you want. In other words, what you are seeing is what you would see if you had an ideal spectrum analyser connected to the output. In this case, when I say ideal I mean it has infinite resolution bandwidth. What I mean by the continuous time spectrum is that all time points are used to compute the spectrum, and so all the little artefacts and imperfections present in the analog signal contribute to the computed spectrum. If you used a DFT or FFT, you would get the sampled spectrum, meaning that the waveform would be sampled at equally spaced points in time and only those values would contribute to the computed spectrum.
What you want depends on what your DAC drives. In other words, your testing procedure should observe the output in the same manner that the surrounding circuitry would. So for example, if you DAC drives a continuous time anti-alias filter, then measuring the continuous-time spectrum makes the most sense. However, if you DAC is followed by discrete time sampling circuits, then generally a DFT of FFT would be more appropriate where you take one sample point per clock cycle, and the sample point falls at the end of the cycle when the signal has settled and all the artefacts and imperfections have settled out. In general the sampled-spectrum should have lower distortion levels than the continuous time spectrum.
The DFT/FFT option is provided by the calculator.
Once you have the spectrum you want, you can compute the various distortion metrics (SNDR, SFDR, THD) by hand or using the calculator.