Arctic Is Planning to Launch First New Thermal Paste in Almost a Decade
Arctic — once known as Arctic Cooling — appears to have a new thermal paste inbound to market. Right now, Arctic sells two types of paste — Arctic MX-2 and Arctic MX-4, with the latter positioned as a more expensive and better-performing product. According to Amazon, there’s a new MX-5 paste on the way, with a claimed lifespan of up to eight years.
How is MX-5 improved over MX-4? No idea, and the Amazon listing doesn’t help a darn bit. The MX-5 claims to be a “carbon-based high performance paste,” which is more or less what MX-4 claims. The lifespan of eight years matches the MX-4 and MX-2.
Comparisons of MX-2 and MX-4 show two things: First, there’s a measurable difference between them, though it varies depending on what kind of cooler you are testing. Second, the gap in measured temperatures isn’t as large as the implied gap based on the specs of the two compounds. MX-2 claims 5.6W/mK, while MX-4 claims 8.5W/mK. These values are known as k-values — they refer to the thermal conductivity of the material, and higher numbers mean a material is more conductive.
This graph from THG compares 85 different thermal compounds. The gap between MX-4 and MX-2 ranges from 0.1C to 0.6C in their various comparisons, but it’s much smaller than the implications of 5.6W/mK versus 8.5W/mK. CoolLab Liquid Ultra claims 38W/mK, and while it obviously outperforms any standard paste, there’s a huge difference between the degree of improvement and the roughly 4.4x difference between MX-4’s thermal conductivity and that of liquid metals. Note: Don’t buy a liquid metal TIM until you’ve read up on how to apply them and what the restrictions are to using them.
The reason the improvement in thermal conductivity doesn’t yield dramatically better temperatures is partly that the thermal paste we apply between heatsink and CPU is one of the last stops on the CPU’s primary heat path. No matter how optimal the material used to transfer heat between the bottom of the heatsink and the top of the CPU heatspreader, there’s still a heatspreader on top of the die in the first place. Below the heatspreader there’s another layer of thermal interface material joining the bottom of the heatspreader and the top of the CPU die. Once we’re inside the die, there are the inevitable thermal conductivity issues with moving heat out of the chip in the first place. All of these contribute to the problem of moving heat out of a CPU.
This is not to imply that thermal paste doesn’t matter — thermal paste matters quite a lot — but the biggest benefit of switching to a better paste these days is typically to quiet a machine, not because you’ll get a dramatic improvement in overclocking performance. While the latter is possible, it typically depends on either a misapplication of decent TIM or the use of a wretched material in the first place. End-users sometimes delid chips to further improve cooling performance, though this only works on a chip that’s used thermal paste instead of solder. It’s technically possible to delid a soldered CPU, but there’s little reason to do so since the solder will be more effective than any alternative an end-user might use instead.
In short, it’s interesting to see Arctic bringing a new product to market, and the overall performance of MX-4 on the chart above suggests there’s room for the company to improve. But the relative paucity of the gap between the best- and worst-performing TIMs, if you don’t count liquid metal, means other factors are stacking up and preventing more dramatic improvements.