GPU memory is divided into two categories: Dedicated and Shared, which combine to make the total GPU memory. Dedicated memory is the VRAM on your card, and is used as the frame buffer as well as assets in use or which will soon be in use, such as textures. Shared memory is comprised of a section of RAM and page file which is used to store assets not in use but may be used again in the future so they are stored for quick access again without having to be regenerated.
Memory clock and memory usage are not related, memory clock is related to utilization and heat, and there is an automatic boost function built into the cards which will increase the speeds while heat and power usage is under a certain limit.
