The world relies on rapid, predictable, and affordable computing performance. However, the digital age continues to grow, pushing extreme data sizes that cripple computer architecture. On average, humanity creates 2.5 quintillion data bytes daily, which is predicted to increase to 463 exabytes by 2025. One can't help but wonder what computer architectures will look like in 50 years or more.
Moore’s Law
Gordon Moore, a co-founder of Intel, wondered just that. He argued that the number of transistors on a chip would roughly double every 18 months (Uddin, 2013). Later, Moore's Law was updated to predict the doubling of transistors every two years. Computer scientists and Engineers used Moore's Law from the 1960s to the 2000s to provide a unified understanding of computer architecture growth, mainly to compare competitors in the industry (Mollick, 2006). Moore did not write a formula for his prediction; however,
is commonly used. In this formula, a is equal to a current position, b is equal to 2 (represents doubling), and t is equal to time. For example, in 1971, the Intel 4004 with a 4-bit, 16-pin chip had a MOS transistor count of 2250. Using Moore's Law,
it was predicted to reach 7200 transistors by 1981.
Predictions In Computer Architecture
Such predictions are directly related to Computer Architecture. For example, assuming the size of a processor and its processing power double every two years, it would be rational to assume that all computer components would be affected. Moore's Law is failing and is anticipated to flatten by 2025 (Shalf, 2020). Several aspects of computer architectures can aid in resisting a failing Moore's Law. The goal is to increase performance via frequency scaling. A higher voltage must be applied to improve frequency, resulting in increased power consumption (Uddin, 2013). The constant power supply required can be complex as mobile device popularity grows. With more power comes more heat. To improve performance, heat must be removed from the chip to avoid damage.
References
Ilatikhameneh, H., Ameen, T., Novakovic, B., Tan, Y., Klimeck, G., & Rahman, R. (2016). Saving Moore’s law down to 1 nm channels with anisotropic effective mass. Scientific Reports, 6(1). https://doi.org/10.1038/srep31501
Lundstrom, M. (2003). Moore's Law Forever? Sciencemag.
Mollick, E. (2006). Establishing Moore’s Law. MIT Sloan School of Management.
Shalf, J. (2020). The Future of Computing Beyond Moore’s Law. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 378(2166), 20190061. https://doi.org/10.1098/rsta.2019.0061
Uddin, I. (2013). Advances in computer architecture. University of Amsterdam, The Netherlands.
The Future of Computer Architecture
The Future of Computer Architecture
The Future of Computer Architecture
The Future of Computer Architecture
The world relies on rapid, predictable, and affordable computing performance. However, the digital age continues to grow, pushing extreme data sizes that cripple computer architecture. On average, humanity creates 2.5 quintillion data bytes daily, which is predicted to increase to 463 exabytes by 2025. One can't help but wonder what computer architectures will look like in 50 years or more.
Moore’s Law
Gordon Moore, a co-founder of Intel, wondered just that. He argued that the number of transistors on a chip would roughly double every 18 months (Uddin, 2013). Later, Moore's Law was updated to predict the doubling of transistors every two years. Computer scientists and Engineers used Moore's Law from the 1960s to the 2000s to provide a unified understanding of computer architecture growth, mainly to compare competitors in the industry (Mollick, 2006). Moore did not write a formula for his prediction; however,
is commonly used. In this formula, a is equal to a current position, b is equal to 2 (represents doubling), and t is equal to time. For example, in 1971, the Intel 4004 with a 4-bit, 16-pin chip had a MOS transistor count of 2250. Using Moore's Law,
it was predicted to reach 7200 transistors by 1981.
Predictions In Computer Architecture
Such predictions are directly related to Computer Architecture. For example, assuming the size of a processor and its processing power double every two years, it would be rational to assume that all computer components would be affected. Moore's Law is failing and is anticipated to flatten by 2025 (Shalf, 2020). Several aspects of computer architectures can aid in resisting a failing Moore's Law. The goal is to increase performance via frequency scaling. A higher voltage must be applied to improve frequency, resulting in increased power consumption (Uddin, 2013). The constant power supply required can be complex as mobile device popularity grows. With more power comes more heat. To improve performance, heat must be removed from the chip to avoid damage.
References
Ilatikhameneh, H., Ameen, T., Novakovic, B., Tan, Y., Klimeck, G., & Rahman, R. (2016). Saving Moore’s law down to 1 nm channels with anisotropic effective mass. Scientific Reports, 6(1). https://doi.org/10.1038/srep31501
Lundstrom, M. (2003). Moore's Law Forever? Sciencemag.
Mollick, E. (2006). Establishing Moore’s Law. MIT Sloan School of Management.
Shalf, J. (2020). The Future of Computing Beyond Moore’s Law. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 378(2166), 20190061. https://doi.org/10.1098/rsta.2019.0061
Uddin, I. (2013). Advances in computer architecture. University of Amsterdam, The Netherlands.