Von Neumann Bottleneck
There has been an improvement in the number of transistors on a chip. More transistors mean that we have increased our ability to store more memory in less physical space. Memory storage is more efficient than ever.
Today, AI and machine learning are being studied. This requires us to store and process a large density of data, which is possible given the environment: processors and storage solutions. Also, Von Neumann Architecture requires us to store data in a separate block, and the processor needs an individual block. These different blocks are connected by buses. Given this architecture, to process these large-density data, the transfer rates must also be at par with the processing speed, maybe even faster. However, over the years, the increase in transfer speedhas only made a few gains.