I debunked that myth from Tony Northrup years ago.

What the P-MP score means is that on any given camera a score that equals the actual pixel count is a theoretically "perfect" lens. When any lens is put on a camera, there will a perceptual "loss" of resolution, or more accurately, it will not allow the camera to display as sharp an image as the "perfect" lens.

For example - the excellent full frame Sigma Art 85mm F1.4 scores as follows:

D810 (36.3mp sensor FX) = 36 P-Mp
D300S (12.3mp DX) = 11 P-Mp
D500 (20.9 mp DX)) = 19 P-Mp

This lens works well with high mp and low mp cameras, regardless of sensor size.

A lens that is less "sharp" like a full frame Sigma 150-500 tests as follows:

D700 (12.3mp FX) = 7 P-Mp
D300S (12.3mp DX) = 5 P-Mp
D500 (20.9 mp DX) = 6 P-Mp

And last but not least - a DX lens like a Sigma 18-200 shows the following:

D300S (12.3mp DX) = 5 P-Mp
D500 (20.9 mp DX) = 6 P-Mp
D5600 (24.2mp DX) = 6 P-Mp

So that lens does not really perform well on high MP sensors - higher resolution sensors are not going to produce perceptually sharper images the lens is clearly the limiting factor.

The inference here is that if you have a 150-500 it will perform best on a D700, returning a P-Mp score of 7 against a possible 12.3 mp or 57% of the sharpness of a perfect lens. The worst case would be a DX camera with a dense sensor, the D500 at 20.9 mp, only shows a P-Mp of 6 against a possible 20.9mp, or 29% of a perfect lens.

The minor differences you are seeing between formats have to do with the print magnification factor. Smaller sensors have to be magnified (enlarged) more so images will appear less sharp.

The correct way to use this P-Mp score is to see which lenses will provide the most resolution on a given body, comparing the P-Mp to the actual Mp.

Your assumption that somehow an FX lens that shows a lower P-Mp score on a DX body isn't taking into account pixel density, and is ignoring the magnification factor.

FWIW, there is no lens that tests at 24 P-Mp on a DX body - the closest is the Sigma Art 85, which shows 20 P-mp on a D5600 or 83% of it's resolution. The same lens on a D810 (with a slightly lower pixel density) is 36 P-Mp - or 99% of it's resolution. You are not likely to see any lens surpass the performance of this lens on any body, and on a 12.3mp DX camera a score of 11 P-Mp or 89% is outstanding, as is the 83% on the D5600.

This is not an easy metric to wrap your head around - I had several lengthy email exchanges with the techs at DXO Mark and this represents how best to make sense of it all.