Now we are going to speak about the CPU, the Central Processing Unit.
The central processing unit or CPU is the component that executes the instructions of the program that is stored in the computer s memory. It is the brain of the computer.
Processors are manufactured in silicon wafers, each one of them can include between five hundred and one thousand processors.
In a modern CPU, we can find more than 2,000 million transistors. And this number is increasing while the miniaturisation process achieves manufacturing of components at more and more reduced scales.
This evolution was outlined by Gordon Moore, co-founder of Intel, in 1965. He predicted that the number of components in a integrated circuit would double each 18 months. This is known as Moore's law and nowadays it is still valid. But it won't last forever. Some studies suggest that the end of this law will arrive in about twenty years.
The processor that we can find in our own computer is developed with a fourteen nanometers technology. In a plate of eighty-two millimetres, we have near 2 billion transistors.
But, how big is fourteen nanometers? Well, you can see the scale in the picture. This size is ten times smaller than a virus, and about fifty times bigger than a single silicon atom.
A computer is able to do very complex tasks, but its CPU only executes simple instructions, it is the software what makes the hard work of converting this complex tasks into these simple instructions. A CPU is only able to get data from the memory, to perform simple arithmetical and logical operations (that are comparisons such as greater than or equal to), to jump to a different part of the program depending on the results and to put data back into the memory, nothing more and nothing less. It has also mechanisms to interrupt a program after receiving an external signal (to be able to cater for requests of devices that need to be attended immediately). How a CPU works? The prefetch unit extracts the next instruction from the memory and the decode unit decodes it to obtain the operation. It later fetches the data needed from the memory and puts it in the CPU s internal memory (that is called the CPU registers). The arithmetic-logic unit is in
charge of performing the operation with the data stored in the registers and obtaining the result, which is returned later to the main memory from the internal registers. After this, the next instruction is loaded from RAM. Sometimes, before loading the new instruction, a jump to another memory location is executed depending on the results of the previous instruction. The control unit organises the complete process synchronised to a central clock.
The performance of the CPU depends on several factors. One of the most relevant and known is the speed, that is measured in hertz (cycles per second). But hertz is not a measure of speed, but a measure of the frequency of the internal clock. Usually more herzts mean more processing speed, but it is not always like this, because there are other mechanisms that also influence processing speed. The CPUs used in nowadays computers have clock frequencies in the gigaherzt range (1 GHz is equivalent to one US billion hertzs) The problem is that, with high frequencies, the temperature rises and with miniaturisation heat dissipation becomes harder, so you will have noticed that several years ago the increments in processor clock speed of new processors got stuck. They are now improving their performance using other techniques as the heat dissipation limit is near of what technology can achieve today.
The clock gives a signal with a given frequency. Each instruction needs a specific number of clock cycles to be executed. For instance, lets assume that an instruction needs 3,5 clock cycles to be executed as in the image. If the computer works at 1 hertz (1 cycle per second), the instruction will need 3,5 seconds to be completed. But, if it works at 2 hertzs (2 cycles per second), it will complete the operation in just 1,75 seconds. The speed of a modern CPU is measured in millions of operations by second. That's why, given the same processor, the more gigahertzs,
the faster it is.
Another important feature is word length. Basically, it is the number of bits that the CPU can receive when accessing the memory. You can imagine it as the number of lanes in a highway. When we increase the number of bits that can be transferred simultaneously, the performance of the CPU improves.
The last characteristic that affects the CPU performance is the number of cores of the processor. A processor with one core only can execute one instruction at a time.
If we add another core, then two instructions of the same or different programs can be executed in parallel, resulting in a processor twice as fast.
Current CPUs for personal computers have four or even six cores.
Cores can be combined with another technology called hyper threading, that allows to execute two different threads of the same program almost in parallel. For example, each tab of a web browser or each avatar of a video game runs in an independent thread.
As result, a single processor can execute 8 instructions in parallel, increasing the performance of the CPU.
In the consumer market, there are two main processor manufacturers, Intel and AMD. Their products are compatible, so we can execute the same instruction set in all of them. For desktop computers, Intel has the i3 i5 and i7 models, with some specific designs for laptops and tablets. The equivalent in AMD are the Athlon processors. Both brands have more powerful CPUs oriented to the workstation and server segments: the Xeon and Opteron respectively.