Technological feat aside:
Revolutionary heat dissipating coating effectively reduces temperatures by more than 10%
78.5C -> 70C = (78.5 - 70) / 78.5 = 0.1082 = 10% right?!
Well, not really. Celsius is an arbitrary temperature scale. The same values on Kelvin would be:
351.65K -> 343.15K = (351.65 - 343.15) / 351.65 = 0.0241 = 2% (???)
So that’s why you shouldn’t do % on temp changes. A more entertaining version: https://www.youtube.com/watch?v=vhkYcO1VxOk&t=374s
I would argue that what makes sense when considering temperature percentages wrt dissipation, is the difference between old and new, divided by the difference between the system at rest and the old temperature.
Which is then a ratio of offsets, rather than a ratio of one offset and a difference with an arbitrarily defined origin.
In this case, it is fair to assume the temperature at rest of the system around 292K, or 19C.
Which would give: (78.5C - 70C) / (78.5C - 19C) = 14.29%, or (351.65K - 343.15K) / (351.65K - 292.15K) = 14.29%.
GamersNexus presents their temperature testing in terms of difference from room temperature, so this is probably how they’d do this comparison.
I’m not sure they’d see a reason to cover ram temperature unless it was approaching actual risk of harm or enabled higher clocks, though. Comparing cases or CPU coolers by temperature makes sense. Comparing GPUs when they’re using the same chip and cooling performance is a big part of the difference between models? Sure. But RAM? Who cares.