²ÝÁñÉçÇø

Moore¡¯s law is based on made by Moore. The doubling every year of the number of transistors on a microchip was extrapolated from observed data.

Over time, the details of Moore¡¯s law were amended to better reflect actual growth of transistor density. The doubling interval was first increased to two years and then decreased to about 18 months. The exponential nature of Moore¡¯s law continued, however, creating decades of significant opportunity for the semiconductor industry. The true exponential nature of Moore¡¯s law is illustrated by the figure below.

A straight-line plot of the logarithm of a function indicates an exponential growth of that function. The figure is courtesy of Intel Corporation, see .

Semiconductor process technology has always increased in complexity. This phenomenon has been the ¡°innovation engine¡± that fuels Moore¡¯s law. In recent times, complexity increases have been accelerating. Transistors are now three-dimensional devices that exhibit counter-intuitive behaviors. The extremely small feature size of advanced process technologies has required multiple exposures (multi-patterning) to accurately reproduce these features on a silicon wafer. This has added substantial complexity to the design process.

All this complexity has essentially ¡°slowed down¡± Moore¡¯s law. Moving to a new process node is still an option, but the extreme complexity and cost of doing so has slowed the pace of migration. Furthermore, each new process node is now delivering less dramatic results in terms of density, performance, and power reduction.  The evolution of semiconductor process technology is reaching molecular limits, and this is slowing the exponential benefits of Moore¡¯s law.