A couple of weeks ago, GamersNexus published a video and article that benchmarked CPU performance across various thermal paste patterns. It’s well established that the best method of applying the compound is to spread it out as thin as possible, so it fills the gaps with something better than air but doesn’t insulate the parts that would naturally make perfect contact. That takes effort, though, and it’s not clear how much that buys you for modern CPUs with integrated heat-spreaders (IHS).
Video credit: GamersNexus
If you’re attaching a heatsink to a GPU or other bare die ASIC? Different story. Their tests are focused on CPUs with heat spreaders.
Long story short? Not so much difference. The “pea sized” method had a little issue because it didn’t fully cover the IHS, but they went on with the tests because it’s supposed to reflect real-world situations, and that was a real-world type of error. Even still, that corresponded to less than a degree Celsius under load (as measured on an Intel Core i7-8086k). The article mentions something about delidding the CPU, although the photos clearly have an IHS (and that’s the point of the test in the first place) so I’m guessing they only took the IHS off temporarily and replaced it.
It’s interesting how close they ended up. I would have thought that 30 minutes of full load would show at least a few degrees of variance, but apparently not, even with a little patch of uncovered space.