February 5 was World Nutella Day (my lack of knowledge of these things is probably why I don’t get invited to the really good parties). To celebrate this event, Cooler Master decided to illustrate the value of the spreadable hazelnut delight by using it as thermal paste. This experiment isn’t without some history. Back in 2002, Dan, of Dan’s Data, compared a variety of thermal goops against Vegemite and toothpaste. The results, hilariously, favored the toothpaste (Vegemite, otherwise known as Satan’s Hellfood, tied with the other thermal compounds).
The video below doesn’t give much context for Nutella’s relative performance but implies that the material is more than capable of keeping a chip nice and cool. So, should you throw out the Arctic Silver in favor of hazelnut?
Understanding thermal paste
The first thing to understand about thermal paste is that, from a physics perspective, it’s inherently a kludge. By the time heat from the CPU core gets to the top of the lid, it’s already passed through the entire chip stack (not counting secondary heat paths), through the solder or thermal paste between the CPU die and the CPU lid, and then passed through the lid material itself. In other words, the heat that makes it to the top of the lid is a fraction of the total heat that can be dissipated.
Once it hits the lid, that heat needs to radiate from the lid to the actual heatsink and fan. Air is a miserable conductor of heat, which is why we put various forms of goop in between the CPU surface and the heatsink. Nutella, likely has a significant amount of oil in it, which is why it conveys heat particularly well, but as it dries, that oil is going to rapidly evaporate. What’s left will be a sort-of Nutella-dust that transmits heat between CPU and heatsink about as well as a wool blanket.
In short, don’t try this at home. But if you have a CPU you don’t care much about, and a list of handy foods you want to experiment with as thermal transfer materials, we promise to publish the results of the first person who conducts a credible test to discover whether ham or turkey makes a better TIM (thermal interface material).
