Types of transistors

As we mentioned in the introduction, transistors are tiny electronic switches that can be turned on or off by applying a voltage to their terminals. By arranging transistors into different configurations, CPU designers can create logic gates and circuits that can perform basic and complex operations based on Boolean algebra. But what are the different types of transistors used in modern processors, and how do they work?

The most common type of transistor used in CPUs is the metal-oxide-semiconductor field-effect transistor (MOSFET), which was invented by Mohamed Atalla and Dawon Kahng at Bell Labs in 1959. A MOSFET consists of four terminals: source, drain, gate, and body. The source and drain are connected to the channel, a thin layer of semiconductor material that allows current to flow through. The gate is separated from the channel by a thin layer of insulating material, usually silicon dioxide. The body is usually connected to the source or to a fixed voltage.

The basic principle of a MOSFET is that by applying a voltage to the gate, we can control the flow of current through the channel. Depending on the type of MOSFET, the channel can be either enhanced or depleted by the gate voltage. There are two main types of MOSFETs used in modern processors: pMOS and nMOS.

pMOS and nMOS

• A pMOS transistor has a channel made of p-type semiconductor material, which means that it has an abundance of holes (positive charge carriers). A pMOS transistor allows current to flow through when the gate is discharged or set low, and blocks current when the gate is charged or set high. This is because when the gate is low, it creates an electric field that attracts electrons from the source and drain, creating an inversion layer that enhances the channel conductivity. When the gate is high, it repels electrons from the channel, creating a depletion layer that reduces the channel conductivity.

• An nMOS transistor has a channel made of n-type semiconductor material, which means that it has an abundance of electrons (negative charge carriers). An nMOS transistor allows current to flow through when the gate is charged or set high, and blocks current when the gate is discharged or set low. This is because when the gate is high, it creates an electric field that attracts holes from the source and drain, creating an inversion layer that enhances the channel conductivity. When the gate is low, it repels holes from the channel, creating a depletion layer that reduces the channel conductivity.

By combining these types of transistors in a complementary way, we can create CMOS (complementary metal-oxide-semiconductor) logic gates and circuits. CMOS logic gates use both pMOS and nMOS transistors to achieve low power consumption and high noise immunity. For example, a CMOS inverter uses a pMOS transistor on top connected to the power line and an nMOS transistor on the bottom connected to ground. The input signal is connected to both gates, and the output signal is taken from between the two transistors. When the input is high, the pMOS transistor is off and the nMOS transistor is on, allowing current to flow from ground to output, making it low. When the input is low, the pMOS transistor is on and the nMOS transistor is off, allowing current to flow from power to output, making it high.

CMOS logic gates can be combined to form more complex logic circuits, such as AND, OR, NOT, NAND, NOR, XOR, etc. These logic circuits can implement Boolean algebra and arithmetic operations. For example, an AND gate can output a 1 only if both its inputs are 1, while an OR gate can output a 1 if either or both of its inputs are 1. By combining multiple logic gates and circuits, CPU designers can implement arithmetic operations such as addition, subtraction, multiplication and division as well as control operations such as branching looping and comparison.

Design challenges

Using transistors in CPUs is not easy. CPU designers have to deal with many problems and choices that affect how well a CPU works. These include:

• Power consumption: How much electricity a CPU uses when it switches transistors on or off. More transistors means more power and more heat.
• Heat dissipation: How well a CPU can get rid of the heat it generates when it uses power. Too much heat can damage a CPU or slow it down.
• Reliability: How likely a CPU is to work correctly for a long time without breaking or making mistakes. Smaller or fewer transistors can make a CPU more fragile or less accurate.
• Performance: How fast and well a CPU can do its tasks. Different tasks and applications need different measures of performance, such as speed, throughput, latency, bandwidth, efficiency, accuracy, etc.
• Scalability: How much a CPU can grow and improve over time. Smaller or more transistors can make a CPU more scalable, but they also face physical and practical limits.
• Cost: How much money it takes to make and use a CPU. Smaller or fewer transistors can make a CPU cheaper, but they also reduce its functionality and performance.

CPU designers have to balance these factors and find the best solutions for different situations.

CPU Design challenges

Future trends

Transistors are not the only way to make CPUs better. CPU designers are always looking for new ideas and technologies to improve their products. In this section, we will mention some of the new things that could change transistors or CPUs in the future, and how they could make them faster, smarter, or cheaper.

• Quantum computing: Quantum computing uses quantum bits (qubits) that can be both 0 and 1 at the same time. Quantum computing could solve some hard problems that normal computers can’t, such as breaking codes, simulating nature, and finding the best solutions. But quantum computing is very hard to do, because qubits are very sensitive and hard to control. Some companies are making quantum processors that use different kinds of qubits, such as superconducting circuits, trapped ions, light particles, etc. These processors are still very new and limited, but they could open a new world of computation in the future.
• Neuromorphic computing: Neuromorphic computing copies the way brains work to do computation. Neuromorphic computing could process complex data better, such as pictures, sounds, words, etc. But neuromorphic computing is very hard to do, because brains are very complex and hard to understand. Some companies are making neuromorphic processors that use different kinds of artificial neurons and synapses, such as CMOS circuits, memristors, spintronics devices etc. These processors are still very new and limited, but they could offer a new way of computing in the future.
• Carbon nanotubes: Carbon nanotubes are tiny tubes of carbon atoms that have amazing properties. Carbon nanotubes could replace or help silicon transistors in CPUs by making them faster, smaller, cooler, and cheaper. But carbon nanotubes are very hard to make and use, because they are very small and hard to handle. Some companies are making carbon nanotube transistors that use different ways to make and connect them, such as chemical vapor deposition dielectrophoresis covalent bonding etc. These transistors are still very new and not ready for market but they could offer a new material for CPU design in the future.
• Graphene: Graphene is a thin sheet of carbon atoms that has amazing properties. Graphene could replace or help silicon transistors in CPUs by making them faster smaller cooler and cheaper. But graphene is very hard to make and use because it is very thin and hard to modify. Some companies are making graphene transistors that use different ways to make and change them such as chemical vapor deposition epitaxial growth chemical doping etc. These transistors are still very new and not ready for market but they could offer a new material for CPU design in the future.

Quantum computing

How many transistors in a CPU?

One of the ways to measure the complexity and capability of a CPU is to count how many transistors it has. Transistors are the basic units of computation that switch on or off to represent and manipulate binary data and logic operations. The more transistors a CPU has, the more functions and operations it can perform.

The number of transistors in a CPU depends on several factors, such as the architecture, the number of cores, the process node, the design, and the generation of the CPU. Different CPUs have different numbers of transistors, ranging from millions to billions.

For example, here are some statistics and examples of how the number of transistors in CPUs has increased over time, following Moore’s law:

• The first microprocessor, Intel 4004 (1971), had 2,300 transistors.
• The first 32-bit microprocessor, Motorola 68000 (1979), had 68,000 transistors.
• The first 64-bit microprocessor, MIPS R4000 (1991), had 1.35 million transistors.
• The first Pentium processor, Intel Pentium (1993), had 3.1 million transistors.
• The first dual-core processor, AMD Athlon 64 X2 (2005), had 233.2 million transistors.
• The first quad-core processor, Intel Core 2 Quad (2006), had 582 million transistors.
• The first six-core processor, Intel Core i7-980X (2010), had 1.17 billion transistors.
• The first eight-core processor, AMD FX-8150 (2011), had 1.2 billion transistors.
• The first ten-core processor, Intel Core i7-6950X (2016), had 3.2 billion transistors.
• The first twelve-core processor, AMD Ryzen Threadripper 1920X (2017), had 9.6 billion transistors.
• The first sixteen-core processor, AMD Ryzen Threadripper 1950X (2017), had 19.2 billion transistors.
• The first eighteen-core processor, Intel Core i9-7980XE (2017), had 6.5 billion transistors.
• The first twenty-eight-core processor, Intel Xeon W-3175X (2019), had 8.6 billion transistors.
• The first thirty-two-core processor, AMD Ryzen Threadripper 2990WX (2018), had 19.2 billion transistors.
• The first sixty-four-core processor, AMD Ryzen Threadripper 3990X (2020), had 39.54 billion transistors.

As we can see, the number of transistors in CPUs has increased exponentially over the years, following Moore’s law. However, this trend may not continue indefinitely, as there are physical and practical limits to how small and how many transistors can be packed on a chip. Therefore, CPU designers may have to look for alternative ways to improve their products in the future, such as using new materials, new structures, new architectures, or new paradigms of computation.

How many transistors in a CPU?

In this article, we have learned about transistors and how they are used in CPUs. We have seen that transistors are the basic units of computation that switch on or off to represent and manipulate binary data and logic operations. We have also seen that transistors are the key to making CPUs powerful and versatile, but they also pose some major challenges and trade-offs that CPU designers have to face when using them. These challenges include power consumption, heat dissipation, reliability, performance, scalability, and cost. We have also mentioned some of the emerging technologies and concepts that could potentially replace or enhance transistors in the future, and how they could offer new possibilities and advantages for CPU architecture and performance. These technologies include quantum computing, neuromorphic computing, carbon nanotubes, graphene, etc. Finally, we have seen how the number of transistors in CPUs has increased exponentially over the years, following Moore’s law. However, this trend may not continue indefinitely, as there are physical and practical limits to how small and how many transistors can be packed on a chip. Therefore, CPU designers may have to look for alternative ways to improve their products in the future.

We hope you have enjoyed this article and learned something new about transistors and CPUs. If you have any questions or comments, please feel free to share them with us. Thank you for reading!