Itt does, actually. It’s a RAM-saving feature. Normally, graphics require a pre-set buffer in RAM equal to the maximum amount it needs. Say, as an example, a video can use anywhere from 2MB-2GB worth of RAM depending on how complicated the individual frame is within the video. The computer has to allocate the full 2GB to accommodate that video for the entire duration of that video playing. With dynamic caching, it figures out on the fly how much RAM is needed and will allocate anywhere from that 2MB-2GB worth of RAM at the time it’s needed and take it away when it’s not needed. Because it’s done in hardware, it’ll be very fast. When the next frame’s needs go down, so does the allocation. That extra RAM is freed up for other things, while under the old standard method, that 2GB of RAM is mostly left empty except for the most extreme parts of the video and goes to waste.
So yeah, it has everything to do with RAM usage, especially since the CPU shares RAM with the GPU. That CPU is more than happy to use that RAM that was freed up. This makes the RAM go further and improves efficiency.