The W/mK thing doesn't work, or holds little water(no pun), as the temperature testing is not standardized.
One manufacturer will test at x degrees C, another will test at y degrees Celsius, and so on. The performance is going to be different across the many samples, so a comparison based on W/mK alone is useless.
@Oxicoi , the pastes that work best/better on gpus are the thick and sticky kinds.
The more fluid pastes are likely to see more collecting around the sides of the die compared to thicker ones due to the glass-like surface of the gpu die and pump out from the cooler mounting pressure.
The Best Pastes list linked by JohnBonhamsGhost still has some use, thanks in part to the Legacy Thermal Paste Testing chart, near the bottom.
Image 4 has a list of several pastes along with a viscosity rating.
-Noctua NT-H2: I can vouch for this one, as I currently use it on a 1080Ti. NT-H1? Ehh, that one can stay with cpus.
-Thermal Grizzly Kryonaut: I can't recommend this one unless the gpu is at least liquid cooled. This one 'burns out' faster the higher your operating temperatures are, which is silly, considering TG advertises it for overclocking.
Higher temperatures(~80C and up) lead to you having to repaste more often, so this stuff costs more than it actually does.
-TG Hydronaut: IMO, the better alternative to Kryonaut, as it doesn't have that silly little weakness.
-Cooler Master Mastergel Maker Nano
-Kingpin KPx
-Gelid GC-Extreme: Was number 4 on the best pastes list.
-Prolimatech PK-3: Was number 1 on the best pastes list.