Who makes the best CPU for gaming? Generally speaking, Intel wins out in pure framerates, though many other factors come into play. With AMD's latest Ryzen 9 3900X and other 3rd Gen CPUs, the battle is far closer, and both companies have their strong suits. The question of AMD vs Intel CPUs (opens in new tab) has become a lot more complex, in other words.
Both companies also make HEDT (High-End Desktop) CPUs for 'enthusiasts,' but with the higher core counts now available in Ryzen CPUs, HEDT is difficult to recommend as a gaming solution. But what about as a pancake cooking solution? Our friends at Tom's Hardware (opens in new tab) recently set out to determine which CPU was best for cooking pancakes. As a result, Tom's might also be looking to buy a new CPU when the Black Friday deals roll around.
You can watch the full video over at Tom's (opens in new tab), but hopefully it goes without saying that doing this on your own home PC is a recipe for disaster. THG increased the throttling temperature to 105C on the Intel Core i9-9980XE, which allowed it to cook a relatively successful pancake in about 18 minutes. It probably needed another 10 minutes to be fully ready, but it was at least viable. It also only crashed once, probably thanks to the VRMs overheating, and a fan pointed at the VRMs appears to have alleviated that problem.
AMD's Threadripper 2990WX didn't fare nearly as well. That might seem like a good thing—having the CPU run cooler is what you'd normally want. There are two problems, however. First, the AMD system (motherboard) didn't allow THG to override the throttling mechanism, so it didn't reach the necessary temperatures to cook. The Threadripper build also crashed and then refused to boot again. I'm a bit concerned that the CPU and/or motherboard may have gone to the great hardware party in the sky, though Tom's doesn't actually say this. Anyway, all Threadripper managed in the same 18 minutes was a gooey mess.
The pancake winner is Intel. Use this information how you will.
I'm thinking we ought to expand on this test with more work put into the cooking/cooling solution next time. Maybe we buy two cheap pans and rig up a way to properly mount them to the CPU socket? Also, we'll need an AMD motherboard that properly overrides the throttling mechanism. Or, you know, we could just use a power meter to see how much juice each CPU uses and declare a virtual winner that way, saving the poor CPUs from an early demise. But then we wouldn't have pancakes.