Computer science is an extremely fascinating field - the development of IT has experienced exponential growth for decades in accordance with Moore's Law. But unfortunately, every exponential ends at infinity and miniaturisation has its physical limits. Do these limits spell the end of classic silicon processors? When can we expect it? And what can you do for your business to avoid losing precious computing power?
Whether we want it or not, the computing power of IT infrastructure is becoming an increasingly critical factor for the success of companies - and ultimately individuals - in the digital era. One of the best-known rules of thumb regarding computing power and IT development is Moore's Law.
Gordon Moore was an American chemist, businessman and inventor - one of the "founding fathers" of the famous Silicon Valley and the equally famous Intel, today's world leader in the production of computer processors of the classic x86 architecture. Interestingly, his father, Walter, was a California sheriff - to whom the somewhat more verbose title of this chapter refers.
In 1965, Gorgon Moore laid down the now-famous Moore's Law in Electronics Magazine, which then read as follows:
"The number of transistors that can be placed on an integrated circuit (e.g., a processor - Ed. note) doubles approximately every 18 months while keeping the price the same."
For simplicity and ease of understanding, we will slightly modify this almost 60-year-old rule (which in computer science is equivalent to almost several geological ages in other - especially humanities - disciplines). Instead of a strict number of transistors, we will consider:
In the current environment, the question is often asked whether Moore's Law is still valid and to what extent it influences the current IT development. For a long time, it seemed to be an infallible truth, even though its author himself originally set its validity at only 10 years. However, the continued rapid increase in the number of transistors on a chip has steadily enabled incredible computing power and IT innovation.
Eventually, the rate of increase in the number of transistors on the chip slowed down over time, doubling approximately once every two years. Even so, Moore's Law is considered a very solid estimate of technological and, in principle, economic developments. In 1995, 30 years after Moore's Law was formulated, its author commented: "This cannot go on forever. The nature of exponential growth is such that we push it beyond its limits and then disaster strikes. In terms of the size of transistors, you can see that we are approaching the size of atoms, which becomes a major barrier. We have another 10 to 20 years before we reach these principle limits."
In recent years, there have been signs that the trend set by Moore's Law may be starting to slow. The reason for this is exactly what Moore himself describes in the previous paragraph - the miniaturization of chips has reached such a level that the gate widths of the most modern transistors are several atoms in size. It won't go much further. Indeed, even these extremely miniature dimensions bring problems in the form of limitations arising from quantum mechanics and other exotic physical phenomena - such as tunneling superbarrier phenomena leading to quantum teleportation of electrons. In human terms, this means potential instability in processors leading to a reduction in computational performance. It also means that achieving the next doubling of the number of transistors on a chip is becoming increasingly difficult.
How to get "in"? There is no need to despair (at least for now). While Moore's Law may not be able to accurately predict future IT developments (the law is currently estimated to be valid for another 30-40 years), computing power is still increasing. Albeit in a slightly different way than Moore predicted at the time. New technologies are being developed to achieve higher computing power, such as quantum processors, optical and neuromorphic systems. These technologies, seeking new approaches and circumventing the physical capabilities of silicon processors, may eventually overcome the traditional limits associated with Moore's Law.
Even with modern and very powerful processors, computing power is often reprehensibly wasted - software slows down faster than the hardware speeds up. In other words, software expands to fill available memory, and programmers often waste hardware resources.
The situation now seems to have two possible solutions - custom software development and the use of cloud services. Both can give you a competitive advantage and enable faster innovation in the digital environment. While software "tailored" to your needs tends to be perfectly tuned and perfectly optimized, cloud IT infrastructure in turn provides immense computing power with dynamic scaling. Plus, it can mean up to 40% cost savings compared to a corporate on-premise IT solution.
Contact us and together we will design a solution that will reliably withstand the ravages of time regardless of Moore's Law and other IT trends.