What is a CPU (Central Pro­cess­ing Unit)?

The CPU is the heart of your computer. It contains the computing power that is essential for the tasks your PC has to handle on a daily basis.

The ab­bre­vi­a­tion CPU refers to the Central Pro­cess­ing Unit of your computer. It is often simply referred to as the processor. The processor is the central hardware component and the heart of your PC. Without it, a computer cannot function at all. This is mainly because the CPU is re­spon­si­ble for all cal­cu­la­tions that are necessary for the PC’s operation.

To un­der­stand the im­por­tance of the CPU, it is necessary to un­der­stand the basic function of a computer. The computer’s cal­cu­la­tions are performed by machine in­struc­tions, which you can think of as in­struc­tions to the processor. All machine in­struc­tions can be rep­re­sent­ed as sequences of ones and zeros, or binary code. This also happens in your computer because the CPU can only process binary in­struc­tions.

There is not one processor, but a whole range of different CPUs. These can be dis­tin­guished primarily by the number of processor cores. However, the field of ap­pli­ca­tion of CPUs also allows dif­fer­en­ti­a­tion between different processor types. Of course, man­u­fac­tur­ers vary, but the market is dominated by two companies in par­tic­u­lar, Intel and AMD.

Single core CPUs only have a single processor core. This means that they can only process one task at a time. They are the oldest CPUs and are now only rarely used because par­al­leliza­tion plays a central role in many modern ap­pli­ca­tions.

The coun­ter­part to single-core proces­sors are multi-core CPUs. They are char­ac­ter­ized by the fact that they have several cores. They often have two or four processor cores (dual or quad core), but a larger number of cores is not uncommon. Proces­sors with a very large number of cores are used es­pe­cial­ly when running servers. The advantage of multi-core proces­sors is obvious: Due to the different, in­de­pen­dent units, they can execute several tasks in parallel and enable smoother and faster work.

If you work with a con­ven­tion­al desktop PC, you are using a desktop CPU. These are the proces­sors installed in PCs. Many modern desktop proces­sors also include an in­te­grat­ed graphics card, which is suf­fi­cient for standard ap­pli­ca­tions.

Basically, there are no major dif­fer­ences between desktop and mobile proces­sors. In most cases, they mainly differ in power con­sump­tion. In general, desktop CPUs are con­sid­ered more powerful than their coun­ter­parts installed in mobile devices like notebooks.

Proces­sors used inside servers differ from CPUs in laptops and PCs. They have a much higher number of cores to ef­fi­cient­ly execute many si­mul­ta­ne­ous op­er­a­tions. In addition, servers usually run around the clock, so the high load can be com­pen­sat­ed with the number of cores.

The processor performs the essential tasks of your computer. They can generally be dis­tin­guished in the following three ways:

1. Pro­cess­ing in­struc­tions. The computing unit is re­spon­si­ble for pro­cess­ing the received in­struc­tions and returning cor­re­spond­ing results.
2. Com­mu­ni­ca­tion with the input and output devices or pe­riph­er­als. The control unit is re­spon­si­ble for this. It also takes care of the in­ter­ac­tion of in­di­vid­ual processor com­po­nents with each other.
3. Data exchange. A con­ven­tion­al PC consists of many com­po­nents, e.g. different types of memory or the graphics card. With its system bus, a processor ensures that data can be sent back and forth between the com­po­nents.

In addition to these central com­po­nents, CPUs can also contain other com­po­nents that have become in­dis­pens­able in modern proces­sors:

• Memory Man­age­ment Unit: The Memory Man­age­ment Unit, or MMU for short, manages access to the computer’s main memory or RAM by trans­lat­ing virtual memory addresses into physical ones.
• Cache: The cache is often a multi-level, fast buffer memory.
• Floating point unit: The floating-point unit is a spe­cial­ized computing unit re­spon­si­ble for handling decimal numbers.

The pro­cess­ing of in­di­vid­ual in­struc­tions within the CPU happens in­cred­i­bly fast. For example, when you press a key on your keyboard, you normally see the cor­re­spond­ing letter on your monitor without delay. Nev­er­the­less, many steps are running in the back­ground for the pro­cess­ing of in­struc­tions to run smoothly. The basic sequence of command pro­cess­ing can be divided into four essential phases:

1. Fetch: First, the address of the next machine in­struc­tion is read from your computer’s memory.
2. Decode: Then the in­struc­tion is decoded, and the cor­re­spond­ing circuits are loaded.
3. Fetch operands: Then all pa­ra­me­ters required for the in­struc­tion are loaded into the registers. The values that are written into the registers can be found either in the main memory, in the working-memory or in cache.
4. Execute: Finally, the command is executed.

These four phases are repeated prac­ti­cal­ly in a con­tin­u­ous loop: as soon as a command has been completed, the next command is selected and processed by the processor. The order in which the commands are executed depends on sched­ul­ing pro­ce­dures. Through ap­pro­pri­ate planning, they ensure that the system behaves in a balanced manner.

How powerful a processor is, depends on several factors. On the one hand, the word width is relevant. This specifies how long a machine word can be. For example, it de­ter­mines how many bits can be read from the main memory at the same time or in which range integer or floating-point numbers can be processed. Most common computers have a word width of 32 or 64 bits.

The number of CPU cores also plays a decisive role when you want to assess the per­for­mance of a processor: The more cores a processor has, the more tasks can be processed in parallel. Load dis­tri­b­u­tion within your system also works better with an in­creas­ing number of cores.

However, it is not only the number of cores that matters. At least as important for the per­for­mance of a CPU is the clock frequency with which the in­di­vid­ual cores work. The clock frequency is specified in Hertz or Gigahertz. Basically, the higher the clock frequency, the more machine in­struc­tions can be processed by the CPU per second.

However, the mainboard’s base clock also plays a role for the clock frequency, which can be set manually in BIOS for some main­boards. Fur­ther­more, the clock frequency can’t be increased ar­bi­trar­i­ly, but is always limited by the CPU tem­per­a­ture. If this increases too much, the processor can be damaged under certain cir­cum­stances. Not least because of this, over­clock­ing CPUs also requires some know-how.

What is more decisive for CPU per­for­mance: the number of cores or the clock frequency? Un­for­tu­nate­ly, there is no clear answer to this question. It not only depends on the ap­pli­ca­tion, but also on the processor itself.

Modern proces­sors are often more efficient in pro­cess­ing in­struc­tions and can therefore also provide the same per­for­mance with a lower clock frequency as older proces­sors with a higher clock frequency. In addition, modern proces­sors often offer the pos­si­bil­i­ty for mul­ti­thread­ing or hy­per­thread­ing, so that several threads can be executed in parallel on one core.

If you run ap­pli­ca­tions on your computer that benefit from multiple cores and par­al­leliza­tion, then it is worth­while to resort to a cor­re­spond­ing­ly high number of processor cores to dis­trib­ute the CPU load in the best possible way. Such ap­pli­ca­tions include the use of virtual machines or rendering. This is because the workload of such programs can be dis­trib­uted very well.

If you mainly use your PC for ap­pli­ca­tions that can’t dis­trib­ute their workload that well, for example computer games, then the clock frequency becomes more decisive.

Modern proces­sors often have an in­tel­li­gent workload dis­tri­b­u­tion across the CPU cores. If the current workload can be ef­fi­cient­ly dis­trib­uted across several cores, this is exactly what is done, and all available cores are used. The in­di­vid­ual cores then run at a lower clock frequency. However, if the use of multiple cores is not sensible or necessary, then the clock frequency of the cores used is increased.