Skip to 0 minutes and 4 secondsYou have just read an article on the advantages in computational complexity that quantum computing seems to offer. We can think of this as the carrot, our reward for building them but there is also a stick. The unrelenting advance of our current computing technology built using transistors made of silicon is slowing down, perhaps even coming to an end. For over half a century, computer hardware has improved at a rapid pace. In fact, we have come to depend on the advances. Every year, we deploy more complex software systems that consume more computing cycles. If we don't have corresponding advances in hardware capabilities, advances in the software that makes our lives better are hard to achieve.
Skip to 0 minutes and 50 secondsThe progressive advance of hardware is known as “Moore's Law.” You may have heard of it. Since Gordon Moore articulated his law in 1965, the semiconductor industry has improved like clockwork, doubling the number of transistors that can be squeezed into a chip every 2 to 3 years. Moore's Law is not an ironclad rule about some physical phenomenon like the laws of thermodynamics. Instead, Moore's Law describes the economic imperative driving improvements in chip manufacturing. It is really about the economic sweet spot of manufacturing. Given current fabrication technology, where is the price per transistor minimized?
Skip to 1 minute and 28 secondsThe death of Moore's Law has been predicted many times over the years and yet, new technologies continue to come online, and research conference present experimental tests of new device and techniques that will appear in products in coming years. But now, we are reaching a level beyond which we cannot go. This is a model of the arrangement of silicon atoms, the same stuff that we make computer chips from. In silicon, the distance between these two atoms is about half a nanometer. This block then represents about one cubic nanometer of silicon. The smallest feature inside a computer chip today is maybe 15 nanometers across or about 10 times the size of this block.
Skip to 2 minutes and 21 secondsWe are building transistors today with a countable number of atoms. This is a mockup of a transistor. Real transistors have several hundred thousand atoms and are very complex structures but we can get an idea of what goes on in a transistor by looking at this mockup of a few hundred atoms. The white block is the channel of a transistor and the orange block is the gate. We have left out many elements including the insulation that appears between the gate and the channel but you can get the idea from looking at this. Silicon is a semiconductor, which means that sometimes a current will flow through it and sometimes not.
Skip to 2 minutes and 59 secondsThe gate controls whether or not electrons can flow from one side to the other. A certain voltage on the gate blocks the flow of electrons, while a different voltage would allow a current of electrons to flow. The structure of the transistor requires special dopant atoms that make it easier or harder for the electrons to flow. In this model, we have colored a couple of the atoms to indicate these dopants. Once the number of atoms in a transistor drops below a certain level, the dopants will no longer be able to influence the electrons properly and we won't be able to build smaller transistors.
Skip to 3 minutes and 38 secondsIt is hard to say exactly when we will reach that final level but certainly transistors that are only a few atoms across would represent the ultimate limit. This doesn't mean that progress will halt entirely. Engineers still have many tricks for making computers better with a fixed number of transistors, and we can continue to increase the number of transistors to a certain extent by stacking them on top of each other. This technique can result in more heat than we can handle, so it still has limitations. So, engineers have been working for several decades to develop what are called "post-silicon" technologies, even though some of them still use silicon as a key material. Quantum computing is one such technology.
Skip to 4 minutes and 23 secondsA quantum computer isn't a magic wand. It can't solve every problem quickly as we will see in detail over the coming weeks but it will be one important tool in our overall effort. Moreover, a lot of what we learn while working on quantum computers can translate fairly directly to advancing other types of computing systems including the concept known as “reversible computing.” The carrot of advancing computation, the stick of working around the impending end of Moore's Law, and there is also one more key motivation for working on quantum computers which you will see in the next article.
Moore's Law: the Technology Incentive
The number of transistors in a CPU continues to increase. Famously, the exponential pace of this growth over the last several decades is called Moore’s Law. What is the current status of Moore’s Law? Will the size of the transistor continue to shrink indefinitely? Can we maintain this pace for the next few years, or even a few decades? The limit of the law may be close; the size of the smallest feature is only a few times the size of a silicon atom.
Even as this growth in density slows, a great deal of room for innovation in computer systems remains. One alternative approach to making large advances in computational capabilities, at least for some problems, is quantum computing.
Subtitles and PDFs
English subtitles are provided for all videos. To enable the subtitles, please hover over the speech bubble icon in the video player, and choose “English”. You can also select “Subtitles off” here to disable it.
Also you can find a PDF version of all the steps including scripts in the DOWNLOADS section of the first step of the each week.
© Keio University