It’s called “3D” because AMD layers the L3 (layer 3) cache on top of the existing CPU components, rather than making the CPU die bigger. By going vertical on the CPU, it’s now a “3d” design. (Other chip makers have done similar things with their own chips, and used the title “3d”)
CPUs keep a copy of some data normally in RAM inside its cache for faster access. This new design allows the cache to be much larger and maintain its speed. Applications that benefit from large caches will run really well on these CPUs, but they will use more power and get hotter. Also, on the CPUs with multiple processing dies only one has the 3D cache on it so the user or app must make sure it runs on the cores on the die that has the 3d cache to benefit from that.
CPU cache is like countertop space in a kitchen. The more counter space you have, the more room you have to place prepped ingredients so they are easy to access while you’re cooking. If you run out of counter space, you’ve got to go and place ingredients on the dining room table in the other room, which takes a lot longer to do each time you have to go place and get it, and you’re not cooking while you’re getting those ingredients.
AMD is adding substantially more counterspace (3DV-cache) into their kitchen (CPU) so the cooks (processor cores) can keep ingredients (temporary data) in quick reach rather than having to store and fetch them in/from the other room (RAM). This amounts to significantly faster performance under certain loads, primarily gaming, since the CPU has what it needs closer to arm’s reach. This does come at a cost of less cocaine for the cooks (slower CPU speeds due to increased heat) but the fact they aren’t waiting for ingredients nearly as much means they are still performing faster.
Latest Answers