The problem is there are different parameters and it's possible to manipulate numbers. As far as I understand, what really matters is transistor density, and it

The problem is there are different parameters and it's possible to manipulate numbers. As far as I understand, what really matters is transistor density, and it obviously can differ for different processes even with same number. Intel of course claim they are better than others on the same numbers (see, for example https://www.extremetech.com/computing/246902-intel-claims-three-year-advantage-10nm-process-wants-change-define-process-nodes https://www.extremetech.com/computing/246902-intel-claims-th... )