Trend that changes everything
The productivity of computers shows impressive and steady growth, doubling every one and a half year. However, few people know that the electrical performance (the number of operations per kilowatt of electricity consumed) also doubles every year and a half from the very beginning of the computer era.
This trend is due to the existence of laptops and smartphones, for which the battery life is crucial. If this law is further observed, in the near future we will see a boom of mobile and miniature devices with very low power consumption, which will greatly increase our ability to collect and process data in real time.
Here is one example of the capabilities of devices with extremely low power consumption: wireless and battery-free sensors that use the energy of electromagnetic radiation on television and radio frequencies, which now fill the air almost anywhere in the Earth. They need only 50 microwatts, and they do not need other power sources.
')
The use of “background” energy sources, such as light, motion, heat, opens up the possibility of widespread use of almost timeless miniature sensors and, consequently, an explosive increase in the amount of available data. It will be possible that Professor Eric Brinjolfson calls the term “ nanodata ” - customized, very detailed data describing the detailed characteristics of individuals, transactions and information flows.
But how long can this trend continue? In 1985, Richard Feynman calculated that the energy efficiency of computers could theoretically be improved at least a hundred billion times, and our data shows that from 1985 to 2009, efficiency increased by about forty thousand times, so we still have everything ahead.
I will give specific examples. A modern MacBook Air, working with the energy efficiency of computers in 1991, would completely discharge the battery in two and a half seconds. The fastest modern supercomputer - Fujitsu K, with a capacity of 10.5 petaflops - consumes 12.7 megawatts, about the same as a small city. But in less than twenty years, a computer of the same capacity will consume as much electricity as a regular toaster. A device comparable to modern laptops will consume almost nothing.
This pattern is observed for all computing devices, but it is not yet known whether the efficiency of data transmission will increase at a comparable pace. Can wireless connectivity be as economical? The transmission speed, frequency range and ways to save energy in idle mode will greatly affect the energy efficiency of mobile devices in general, but improving processors will undoubtedly push this area, because only by improving the characteristics of wireless interfaces you can use the full potential of an economical stuffing.
The long-term trend of increasing energy efficiency of computing will revolutionize the way information is collected and processed. The Internet of Things will become a reality and this will have profound implications for business and society as a whole. This will allow more precise control of production processes, faster and easier to see the consequences of our actions, and, ultimately, change social institutions and business models. This will help to test the assumptions and hypotheses on real-time information in real time, that is, to rely more on reliable experimental data from the real world.
Previously, the efforts of scientists and engineers were aimed primarily at improving the computing power of processors. This is very important today. But the continuous improvement of energy efficiency creates new challenges - now it’s equally important to make the whole device consume as little energy as possible. Over the next ten years, the power consumption of computers should be reduced by about a hundred times. What new products and devices will be created thanks to this? What other innovations will be needed to make the most of these opportunities?
About the author: Jonathan Kumi - a scientist, writer, entrepreneur, professor at Stanford University.
This trend is due to the existence of laptops and smartphones, for which the battery life is crucial. If this law is further observed, in the near future we will see a boom of mobile and miniature devices with very low power consumption, which will greatly increase our ability to collect and process data in real time.
Here is one example of the capabilities of devices with extremely low power consumption: wireless and battery-free sensors that use the energy of electromagnetic radiation on television and radio frequencies, which now fill the air almost anywhere in the Earth. They need only 50 microwatts, and they do not need other power sources.
')
The use of “background” energy sources, such as light, motion, heat, opens up the possibility of widespread use of almost timeless miniature sensors and, consequently, an explosive increase in the amount of available data. It will be possible that Professor Eric Brinjolfson calls the term “ nanodata ” - customized, very detailed data describing the detailed characteristics of individuals, transactions and information flows.
But how long can this trend continue? In 1985, Richard Feynman calculated that the energy efficiency of computers could theoretically be improved at least a hundred billion times, and our data shows that from 1985 to 2009, efficiency increased by about forty thousand times, so we still have everything ahead.
I will give specific examples. A modern MacBook Air, working with the energy efficiency of computers in 1991, would completely discharge the battery in two and a half seconds. The fastest modern supercomputer - Fujitsu K, with a capacity of 10.5 petaflops - consumes 12.7 megawatts, about the same as a small city. But in less than twenty years, a computer of the same capacity will consume as much electricity as a regular toaster. A device comparable to modern laptops will consume almost nothing.
This pattern is observed for all computing devices, but it is not yet known whether the efficiency of data transmission will increase at a comparable pace. Can wireless connectivity be as economical? The transmission speed, frequency range and ways to save energy in idle mode will greatly affect the energy efficiency of mobile devices in general, but improving processors will undoubtedly push this area, because only by improving the characteristics of wireless interfaces you can use the full potential of an economical stuffing.
The long-term trend of increasing energy efficiency of computing will revolutionize the way information is collected and processed. The Internet of Things will become a reality and this will have profound implications for business and society as a whole. This will allow more precise control of production processes, faster and easier to see the consequences of our actions, and, ultimately, change social institutions and business models. This will help to test the assumptions and hypotheses on real-time information in real time, that is, to rely more on reliable experimental data from the real world.
Previously, the efforts of scientists and engineers were aimed primarily at improving the computing power of processors. This is very important today. But the continuous improvement of energy efficiency creates new challenges - now it’s equally important to make the whole device consume as little energy as possible. Over the next ten years, the power consumption of computers should be reduced by about a hundred times. What new products and devices will be created thanks to this? What other innovations will be needed to make the most of these opportunities?
About the author: Jonathan Kumi - a scientist, writer, entrepreneur, professor at Stanford University.
Source: https://habr.com/ru/post/143416/
All Articles