How many Transistors in a CPU?

Transistors are the basic building blocks that regulate the operation of computers, mobile phones, and all other modern electronic circuits and is the basic unit of the CPU. Transistors can be packaged independently or in a small area. Integrated circuits can accommodate 100 million or more transistors.
Looking inside the i9-9900K: A single 14nm++ Trigate Transistor

Catching a single Transistor

Abstract

The transistor in electronic components is a semiconductor device, which is often used in amplifiers or electronically controlled switches. Transistors are the basic building blocks that regulate the operation of computers, mobile phones, and all other modern electronic circuits. At the same time, the transistor is the basic unit of the CPU. Transistors can be packaged independently or in a small area. Integrated circuits can accommodate 100 million or more transistors. The CPU mainly includes a logic operation unit, a control unit, and a storage unit. This article will detail how many transistors in a CPU and how the CPU works.

I Introduction to Transistor

Transistor generally refers to all single components based on semiconductor materials, including diodes, triode, field-effect transistors, thyristors, etc. Transistors sometimes refer to triode. Transistors are mainly divided into two categories: bipolar transistors (BJT) and field-effect transistors (FET). The transistor has three poles; the three poles of the bipolar transistor are composed of N-type and P-type emitters, bases and collectors; the three poles of the field-effect transistor are Source, Gate and Drain.

triode

triode

Transistors are the basic building blocks that regulate the operation of computers, cell phones, and all other modern electronic circuits. Due to its fast response and high accuracy, the transistor can be used for a variety of digital and analog functions, including amplification, switching, voltage regulation, signal modulation, and oscillator. Transistors can be packaged individually or in a very small area.

In December 1947, a research team composed of Shockley, Badin, and Bratton of Bell Labs in the United States developed a point-contact germanium transistor. The advent of transistors is a major invention in the 20th century. After the emergence of transistors, people can use a small, low-power electronic device to replace the bulky, power-consuming tube. The invention of the transistor blew the horn for the birth of integrated circuits. In the first 10 years of the 20th century, semiconductor materials have been used in communication systems. In the first half of the 20th century, the ore radio, which was widely popular among radio enthusiasts, used ore as a semiconductor material for detection. The electrical characteristics of semiconductors have also been used in telephone systems.

II Number of CPU Transistors in Previous Generations

 

Moore's Lawvs. Intel Microprocessor Density

Moore's Law vs. Intel Microprocessor Density

Moore's Law was proposed by Gordon Moore, one of the founders of Intel. When the price is unchanged, the number of components that can be accommodated on the integrated circuit will double every 18-24 months, and the performance will double. In other words, the performance of a computer that can be bought for every dollar will more than double every 18-24 months. This law reveals the speed of information technology progress. Although this trend has continued for more than half a century, Moore's Law should still be considered observation or speculation, not a physical or natural law.

(1) February 1999: Intel released the Pentium III processor. Pentium III is a 1 × 1 square silicon with 9.5 million transistors, manufactured using Intel’s 0.25-micron process technology.

(2) In 2000, Pentium 4 Willamette, the production process is 180nm, the number of CPU transistors is 42 million.

Intel Pentium 4

Intel Pentium 4

(3) January 2002: The Intel Pentium 4 processor is launched, and high-performance desktop computers can achieve 2.2 billion cycles per second. It is produced using Intel's 0.13-micron process technology and contains 55 million transistors.

(4) March 12, 2003: The Intel Centrino mobile technology platform was born on notebook computers, including Intel’s latest mobile processor, the Pentium M processor. The processor is based on a new mobile-optimized micro-architecture, produced using Intel ’s 0.13-micron process technology. It contains 77 million transistors.

(5) May 26, 2005: Intel's first mainstream dual-core processor, the Intel Pentium D processor with 229,999,999 transistors using Intel's leading 90 nm process technology.

(6) July 27, 2006: The Intel Core 2 dual-core processor was born. The processor contains more than 290 million transistors, uses Intel's 65-nanometer process technology. It is produced in several of the world's most advanced laboratories.

(7) January 8, 2007: To expand sales of quad-core PCs to mainstream buyers, Intel released the Intel Core 2 quad-core processor and two other quad-core server processors for desktop computers with a processing power of 65 nanometers. The Intel Core 2 quad-core processor contains more than 580 million transistors.

Intel Core 2 quad-core processor

Intel Core 2 quad-core processor

(8) The Corei7 980X launched in 2010, the production process is 32 nm, and the number of transistors is 11,699,999,999.

(9) The Corei7 4960X launched in 2013 has a manufacturing process of 22 nm and a transistor count of 1.86 billion.

III Why are the CPUs with more Transistors stronger?

The CPU is like a large factory for storing switches. Each transistor is a switch, that is, 0 when turned off and 1 when turned on. The more transistors, the more switches. When you deal with the same problem, the more routes you choose, the more looped lines.

Similarly, the more transistors a CPU has, the more branches current flows in a unit of time. From a macro perspective, the more data you can process on the CPU, the faster the machine.

The CPU mainly includes a logic operation unit, a control unit, and a storage unit. There are some registers in the logic operation and control unit. These registers are used for the temporary storage of data in the process of data processing by the CPU.

CPU manufacturing is a very high-precision process, which must be completed by means of machinery. The transistors cut from the wafer will be printed on the CPU base by machinery. Each printed base will be checked to eliminate defective products. And then add the shell interface to become our common processor product.

IV How does the CPU work?

As we all know, the CPU is the "heart" of the computer and the core of the entire microcomputer system. Therefore, it is often synonymous with various grades of microcomputers, such as 286, 386, 486, Pentium, PII, K6 to today's PIII, P4, K7, and so on. Looking back at the history of CPU development, the CPU has been greatly improved in manufacturing technology. It mainly reflected in the integration of more and more electronic components. From the beginning of integrating thousands of transistors to the current millions and tens of millions of transistors. How do they process data?

1. The Original Working Model of the CPU

 

Before understanding the working principle of the CPU, let's talk briefly about how the CPU is produced. The CPU is manufactured on pure silicon material. A CPU chip contains millions of delicate transistors. People use chemical methods to etch or photoetch transistors on a silicon wafer. Therefore, the CPU is composed of transistors. Simply put, transistors are miniature electronic switches. They are the cornerstone of building a CPU. You can think of a transistor as a light switch. They have an operating bit, which represents two states: ON and OFF. This on and off is equivalent to the connection and disconnection of the transistor. These two states correspond to the basic states "0" and "1" in binary. In this way, the computer has the ability to process information. But don't think that the principle of the simple "0" and "1" two-state transistors is very simple. In fact, their development is obtained after years of hard research by scientists. Before transistors, computers relied on slow, inefficient vacuum tubes and mechanical switches to process information. Later, scientists put two crystals into a silicon crystal, so that the first integrated circuit was created.

Seeing this, you must be wondering, how do transistors use the two electronic signals "0" and "1" to execute instructions and process data? In fact, all electronic devices have their own circuits and switches. The flow or disconnection of electrons in the circuit is completely controlled by the switch. If you set the switch to OFF, the electron will stop flowing. If you set it to ON, electronics will continue to flow. The switching of ON and OFF of the transistor is only controlled by electronic signals, so we can call the transistor a binary device. In this way, the ON state of the transistor is represented by "1", and the OFF state is represented by "0", which can form the simplest binary number. The special order and pattern of multiple "1" and "0" generated by many transistors can represent different situations, which are defined as letters, numbers, colors, and graphics. For example, 1 in decimal digits is also "1" in binary mode, 2 is "10" in binary mode, 3 is "11", 4 is "100", 5 is "101", and 6 is " 110 "and so on,  which constitutes the binary language and data used in computer work. Groups of transistors can be combined to store numerical values, as well as to perform logical and digital operations. 

2. The Internal Structure of the CPU

 

 transistors in CPU

transistors in CPU

Now we already know roughly what the CPU is responsible for, but which components are responsible for processing data and executing programs?

1) ALU (Arithmetic Logic Unit)

ALU is the core of the arithmetic unit. It is based on a full adder, supplemented by a circuit composed of a shift register and corresponding control logic. Under the action of the control signal, it can complete four operations of addition, subtraction, multiplication, and division, and various logical operations. As just mentioned, this is equivalent to the production line in the factory, which is responsible for calculating the data.

2) RS (Register Set or Registers)

RS is essentially a place where data is temporarily stored in the CPU. It stores the data waiting to be processed or the data that has been processed. The time for the CPU to access the register is shorter than the time to access the memory. The use of registers can reduce the number of times the CPU accesses memory, thereby improving the CPU's working speed. However, due to the limitation of the chip area and integration, the capacity of the register set cannot be large. The register set can be divided into special registers and general registers. The general-purpose register is widely used and can be specified by the programmer. The number of general-purpose registers varies by the microprocessor.

3) Control Unit 

Just like the logistics distribution department of the factory, the control unit is the command and control center of the entire CPU. The control unit consists of three components: the instruction register IR, the instruction decoder ID, and the operation controller OC. It is extremely important to coordinate the orderly work of the entire computer. According to the user's pre-programmed program, the control unit takes each instruction from the register. In turn, the register puts it in the instruction register IR, determines what operation should be performed by instruction decoding (analysis), and then operates the controller OC in accordance with the determined timing, sends micro-operation control signals to the corresponding parts. The operation controller OC mainly includes control logic such as beat pulse generator, control matrix, clock pulse generator, reset circuit, and start-stop circuit.

4) Bus

Just like the communication channel between various parts in the factory, the bus is actually a set of wires, a collection of various common signal lines. The bus used as a "highway" for the common use of all components in the computer to transmit information. The bus directly connected to the CPU may be called a local bus, including DB (Data Bus), AB (Address Bus), CB (Control Bus). Among them, the data bus is used to transmit data information; the address bus is used to transmit address information issued by the CPU; the control bus is used to transmit control signals, timing signals, and status information.

3. CPU Workflow

 

The CPU, which is composed of transistors, is the core of processing data and executing programs, namely the Central Processing Unit. First of all, the internal structure of the CPU can be divided into a control unit, logical operation unit, and storage unit (including internal bus and buffer) three parts. The working principle of the CPU is like a factory's processing of products: raw materials (program instructions) that enter the factory are dispatched by the material distribution department (control unit) and are sent to the production line (logical operation unit) to produce finished products (processing Data), and then stored in the warehouse (storage unit), and finally wait to be sold on the market (for use by the application). In this process, we noticed that from the control unit, the CPU started the formal work. The intermediate process is to perform the arithmetic processing through the logical operation unit, and the handover to the storage unit represents the end of the work.

4. Data and Instructions

 

Now, let's see how the data runs in the CPU. We know that data flows from the input device through the memory and waits for processing by the CPU. The information to be processed is stored in bytes, that is, 8-bit binary numbers or 8 bits are stored as a unit. These pieces of information can be data or instruction. Data can be characters, numbers, or colors in binary representation. The instructions tell the CPU what operations to perform on the data, such as completing addition, subtraction, or shift operations. We assume that the data in memory is the simplest original data. First, the instruction pointer will notify the CPU to place the instruction to be executed in a storage location in memory. Because each storage unit in the memory has a number. The data can be taken out according to these addresses and sent to the control unit through the address bus. The instruction decoder takes the instruction from the instruction register IR and translates it into the executable form for CPU. And then CPU decides what necessary operations are needed to complete the instruction. It will tell the arithmetic logic unit (ALU) when to calculate, tell the instruction reader when to get the value, tell the instruction decoder when to translate the instruction, etc. If the data is sent to the arithmetic logic unit, the data will perform the arithmetic operations specified in the instructions and various other operations. When the data processing is completed, it will return to the register, and continue to run the data through different instructions, or send it to the data buffer through the DB bus.

Basically, the CPU performs three basic tasks: reading data, processing data, and writing data to memory. But under normal circumstances, an instruction can contain many operations performed in a clear order. The job of the CPU is to execute these instructions. After completing an instruction, the CPU's control unit will tell the instruction reader to read the next instruction from the memory and execute. This process repeats continuously and quickly, executing one instruction after another, producing the results you see on the display. While processing so many instructions and data, there will definitely be chaotic processing due to the data transfer time difference and the CPU processing time difference. In order to ensure that every operation occurs on time, the CPU needs a clock, which controls every action performed by the CPU. The clock is like a metronome. It continuously pulses and determines the pace and processing time of the CPU. This is the nominal speed of the CPU we are familiar with, also known as the main frequency. The higher the main frequency value, the faster the CPU works.

5. How to improve CPU work efficiency?

 

Since the main job of the CPU is to execute instructions and process data, work efficiency will become the main content of the CPU. CPU manufacturers also try their best to use the CPU to process data faster. According to the internal calculation structure of the CPU, some manufacturers add another arithmetic logic unit (ALU) in the CPU or set up another floating-point operation unit (FPU). It greatly speeds up the speed of data operations. In terms of execution efficiency, some manufacturers increase the execution speed of instructions by pipelining or executing instructions in parallel. As we mentioned earlier, the execution of instructions requires many independent operations, such as instruction fetching and decoding. Initially, the CPU must execute the previous instruction before executing the next instruction, but now the distributed circuits perform their operations. That is to say when this part of the circuit has completed one job, the second job immediately occupies the circuit, which greatly increases the efficiency of execution. In addition, to make the connection between instructions more accurate, the current CPU usually uses a variety of prediction methods to control instructions to execute more efficiently.