A tester at Tom's Hardware has just published the results of a huge 85 product test of thermal compounds to measure the impact they have on cooling performance and usability. Liquid metal was found to be the most efficient, but there was a smaller difference across the board than we would have expected.
https://www.notebookcheck.net/Liquid-metal-found-to-be-the-most-effective-in-a-massive-thermal-comound-test-by-Tom-s-Hardware.240236.0.html
Conductonaut can drop temps more than 10C on the CPU compared to Kryonaut. These temps drops can only be seen on highly overclocked laptops (like 7820HK at 4.5 ghz). At lower power draws or at stock, there isn't much difference. On laptop video cards, there's barely any difference, due to much higher surface area on the chip, compared to BGA or delidded (under the IHS) CPU's. Desktop 1080 GTX's should show a bigger improvement than laptop 1080's.
My own testing shows the same, I tested Thermal Grizzly Kryonaut (non-conductive) vs Coollaboratory Liquid Ultra (liquid metal) on my desktop GPU - GTX 1070, and the liquid metal was only 1 degC cooler than the Kryonaut. Not worth using in my opinion unless you use it underneath the IHS of a delidded CPU.
Falkentyne - That was the result I was expecting, a greater separation as the voltage was increased.
my own testing shows same - I feel the same as you, I'm not normally pushing the thermal load to the point where I need a liquid metal. Therefore I feel much more comfortable with non-conductive pastes.