Moore's Law revisited through Intel chip density.

Area of Science:

• Semiconductor physics
• Computer engineering
• Materials science

Background:

• Gordon Moore's observation posits exponential growth in transistor count per integrated circuit, doubling every 12-24 months.
• Debate exists on whether simple exponential growth accurately models computer processor evolution dynamics.
• Processor evolution involves both increased transistor integration and miniaturization.

Purpose of the Study:

• To analyze the historical evolution of transistor density in computer processors.
• To determine if transistor density follows exponential or other growth models.
• To identify patterns and potential future limits in processor manufacturing.

Main Methods:

• Analysis of transistor density in Intel processors from 1959 to 2013.
• Modeling processor evolution using a single measure: transistors per unit area (chip density).
• Fitting the data to growth curves to identify characteristic times and patterns.

Main Results:

• Transistor density growth is consistent with a biphasic sigmoidal curve, not simple exponential growth.
• The model indicates characteristic times of 9.5 years for the sigmoidal curve.
• Six distinct waves of transistor density increase were identified, each with rapid growth followed by stagnation.

Conclusions:

• Processor evolution is characterized by distinct phases of rapid advancement and slower growth.
• Understanding these biphasic sigmoidal patterns provides insight into manufacturing processes.
• This analysis highlights potential future limitations and areas for overcoming them in processor development.