1)
When computing the Gamut Coverage of a profile generated after a rec709 virtual probe to UHD rec 2020 a I have got the following:

That depends on the size of the chosen profile. Using a 3x3x3 profile I get a correct 54%. In the above case I have used a 21x21x21 profile (the max in built in).
I have attached it for replication. Anyhow already using a 4x4X4 profile I get 100%.

2)
Also noted that gamut coverage does not depend on the target max luma.
E.g. 5x5x5 rc709 profile at 100 Nits has 99% coverage regardless the Target Luma max. I guess it is intended (i.e. it makes little sense comparing gamut with different max lum). Is that correct?

Attached file: 21X21X21 rec709 profile

Seems you have found a bug with the virtual probe's perfect data sets.
Have you tested real profiles?

And yes, the value reported is 'Gamut' coverage.
Both the profile and the target colour space are normalised, as that is what gives the 'Gamut' coverage.

Steve

With real profile no issue.
If you move one of the points of the perfect profile, it works fine.

I have edited the topic heading.

Steve