The server CPU is the most important part of your server. When you buy any digital processing tool (including servers, storage, firewalls, personal computers, laptops, tablets, mobile phones, and even smart watches), you must know the name “processor” or so-called Central Processing Unit. You have heard of CPU). The server CPU is the main “brain” or “center” of decisions on your computer and manages all the instructions performed on your computer. Also, all information processing your server or computer must do is done in this section.
A better understanding of zero and one
When you set up a server, whether you want a simple image to be shown or a huge information processing, everything must be done according to digital rules and related processes.
Server processors are built on the principles of electronic engineering and very complex materials. Still, at the Core of all of them, everything is based on the principles of zero and one and logical functions. There are simple structural principles in every process; by understanding them, we can get to know the other functions of a computer (whether it is a smartwatch or big super chips).
To better understand how processors work, read the rest of this article carefully to get acceptable results.
Principles of processor work
If we look at a processor, we realize that processors are very complex. But the main function of any processor is performed by its silicon parts. In silicon parts, everything is measured in microns one thousandthshs of a millimeter). Things that may not be possible to see with the naked eye. That’s why when you look at a processor, you see billions of tiny pieces stacked together.
As we will see below, each of these small parts (read transistor) is responsible for passing specific signals. The passing of these signals (in the form of zeros and ones) and the arrangement of these signatories from the processes of our world. In the world of zero and one (or binary), everything is expressed based on zero signals. One and the code processed in the processors’ transistorsEveryry process has a large complex of transistors whose task is to process these zero and one signalsIncreasingng the number of transistors in server CPU s does not necessarily mean increasing the processing speed. However, in each generation, with the increase in the number of transistors in the server CPUs, we saw an increase in their speed as well. To find a higher speed in processing CPUs, we must use other structures that help to better manage processes and data flow. We will examine these issues in the rest of this article.
To better understand what increases the CPU speed of servers and computers, we must have a correct understanding of how instructions are processed (Instruction set). Each instruction is a series of logical arrangements of zeros and ones that tell the processor what processes to perform and what results to return.
In server CPUs and computers, we have different sets of instructions. Instructions are named x86 or x32 or whatever, based on their logical architecture. We will provide you with explanations about the reason for this name.
Understanding the concept of Core, CAC, he, and graphics in CPToto understand how a server CPU accommodates a large number of processors and manages them, we must have a proper understanding of the structure of the processors.
List of articles
Basic understanding of CPU core
In CPUs, transistors do not work by themselves and independently of each other. Rather, each group of transistors is placed in a set called “Core.” Each processor has a certain number of cores (always an even number) depending on its structure. Each Core receives a series of instructions and returns the result after processing. All the processes that are done in CPUs happen in the cores. In today’s server CPUs, we see several cores that perform virtual processes to maximize the number of cores in a processor. We use the concepts of hyper-threading and simultaneous processing.
Parallel processing means each processor’s core processes information while other cores in the CPU work. This work speeds up the processor and increases its efficiency rate of a processor. Parallel processing allows us to run several software simultaneously on a computer or server. Also, many activities that we don’t see are done behind the scenes.
Threaded processing is a technique to better use the Power of each Core. Each Core has specific times to receive, process, and send responses. In thread processing we use a scheduler to send information to the Core continuously. This can increase the processing speed by two times or more (depending on the type of programming and instructions). Normally, a processor that has, for example, eight cores, with the use of thread processing, its Power will be equal to sixteen cores. We will give you more explanations about this issue in the rest of the article.
A basic understanding of the concept of cache
The next concept that helps increase the Power of server CPUs is Caching. Cache means to receive and store information quickly for processing. In the past, the closest memory to a processor was RAM. RAMs are very fast memories and hold a large amount of data (up to 8 GB and even more). RAMs have a much higher speed than other memories such as hard disks or solid state memories (SSD). However, even these fast RAMs are too slow to keep up with the server’s CPU. In recent years, the use of cache memories in processors has flourished, and now we see the use of three-level caches in new processors. New caches give the CPU cores the ability to quickly fetch their information and save the result immediately. Then this information is quickly sent to the RAM and finally to the relevant hardware. The more cache space in a processor, the faster its processing speed. Cache space is often expressed in megabytes (MB).
List of articles
Basic understanding of CPU graphics concept
Another concept that should be mentioned in the topic of server CPU is the use of graphic processing units. As mentioned, every CPU server or computer uses cores and transistors for its processes. In old processors, all sets of instructions were sent to the cores, and then the response of the processes was received and sent to the relevant hardware. This is great for computing and processing tasks, but it was a problem regarding media playback. For this reason, in new CPUs, for better processing of graphic instructions (such as displaying graphic interface, videos, rendering games, etc.), they use an authorized component called “Graphics processing unit – Graphics processing unit”.
This unit is designed to process graphics commands and work better with graphics hardware. A graphics processing unit (GPU) is required in new generation processors. If you plan to do graphic work (such as using Photoshop or 3DMax software, or even a simple Windows operating system), having a GPU will greatly help increase the CPU efficiency of your server.
Installing the CPU on the chassis and motherboard
In order for the server CPU to be placed in the chassis and motherboard of the server, an interface called “Socket” must be used. Sockets are not a separate part of the motherboard. These sockets are a communication bridge between the processor and the motherboard.
Each socket is built based on a specific architecture. This architecture allows communication with specific processors to each socket. A certain number of lenses (lanes) are used in each socket. Lenses can be simple protrusions, needle-shaped, or metal contact plates. The information goes to the processor through the lenses and returns to the motherboard.
Depending on the type of processor and chassis, a certain number of lenses are used in each socket. To better understand how lenses work, we need to understand how processors work and process information. For this purpose, read the next part of this article.
How does the server CPU work?
So far in the article, we have learned about the different hardware parts of a server or computer CPU. But maybe you have also been asked how a processor can do processing tasks. To better understand the operation of a CPU in a server or computer, you must understand some concepts.
List of articles
The concept of a cycle in computer
Both server and computers are made of many circuits. A large amount of information is moving in these circuits. By the time you read this article, there is a lot of information to be processed regularly. For example, your monitor image now refreshes sixty times per second. (Of course, you won’t notice this value, because your eye doesn’t understand the speed of change for more than 15 seconds). Also, every time you move this page, a large amount of information has to be processed for you to understand an organized and coherent graphic picture of what I have written for you.
All these processes are done using different processes in a computer. For the process of transferring and processing data in a processor to be done correctly, we need to use a concept called “Cycle.” Each cycle is a certain time unit.
Every computer’s cycles are determined using a microchip with the Power to show time on a millisecond scale (one-thousandth of a second). (Of course, in current computers, time intervals less than a millionth of a second are also used). Each cycle in our computer is reported in the form of signals with different parts. As a result, all the hardware of our computers is coordinated with each other and sends or receives information based on it. This way of working will be the same on servers and smaller computers.
To express the concept of a work cycle, they use expressions like Clock. The higher the clock speed of a processor, the faster the processor receives information in smaller time intervals, which can help increase its processing speed. And vice versa, the lower the Clock is, the processor performs the processing in higher time intervals, and as a result, its processing speed will be lower.
After the data movement is determined based on specific time cycles, each part must send its desired data to the processor to decide based on different instructions.
The concept of frequency
Each CPU can process a set of information in each time interval. For ease of work and creating a standard, processors use the “number of instructions processed per second” unit. The higher the number of instructions processed per second, the faster your processor will be. Today’s processors can process billions of instructions per second. Giga (G) units are used to show the number of commands processed in one second. For example, most current processors have a frequency speed equal to 2G/s. That means processing two billion processing instructions per second.
In multi-core and multi-threaded CPUs, the processing load is distributed among different cores, and the mentioned speed may not necessarily be accessible to the processor. However, the higher the frequency speed of a processor, the higher your server and computer speed.
It should be noted here that the higher speed in servers or motherboards is not necessarily dependent on the processing speed of the processor, and several other factors must work correctly to get the desired result.
Motherboard structure
All components installed on a server or computer must send this information to the processor to process their information. Sending information from all hardware to the processor can make our work difficult. For this reason, before the information flows to the server CPU, various parts are used that we should check in place, but three parts are very important for us:
- Northbridge: This bridge organizes all incoming information. When you press a key on the computer, the information from the keyboard is sent to the Northbridge. Here the digital codes and the processing channel are specified.
- RAM Memory: Here, the digital information of the keyboard is received from the north bridge; also, the instructions related to the keyboard are called and wait for processing.
- South Bridge: After the information goes to the processor, it is processed, and the outputs are delivered to the RAM again. The RAM sends this information to the south bridge, and the south bridge sends the information based on the desired output. Here, for example, if you have typed the letter “b” from the keyboard, the letter “b” will be displayed on the screen.
So far, we have understood the input and output of information to the CPU of the server and computer, but what happens inside the CPU itself and how is the information processed? Read the next part of this article to understand this better.
CPU work steps
To perform a processing process, several important steps must be performed in a processor so that the processing results are delivered to us. For this purpose, a processor must go through several general steps, which are as follows:
- Fetch stage: In this stage, information is directed from the side of the RAM to the cache memories and then to the processor cores. Here the information is encrypted and can only be understood by the caches and fetching cores in the CPU.
- Decoding stage: In this stage, the fetched packet or the term “Word” is decoded. This word contains instructions that the CPU core must process. These include content information and processing instructions for them. Codes here are a set of zeros and ones and logical relationships between them.
- Processing stage: This information is processed based on the content and instructions, and the processing results are ready to be sent. In this part, the commands are processed by ALU or Arithmetic and Logic Unit. What is processed here is nothing like what is shown to you. Everything is based on logical equations (AND, OR and others) and zero and one. Information is processed in this way at the level of transistors.
- Rewrite stage: In this stage, the processor does its work and sends its final information to internal memories and then to external memories such as RAM memory.
Compatibility of CPU architecture
One of the most important points in choosing the right CPU is matching the CPU with your server architecture.
Each server uses a specific chassis, motherboard, hardware system, and software firmware to communicate with those hardwares. If we want a suitable CPU for our server, we must pay attention to the motherboard’s specifications and the hardware system (including the socket, clock speed, and RAM processing power). The most important things to pay attention to here are:
- Processor socket: Each hardware has a specific match with sockets. There are different types of sockets in the market. When buying a processor, pay attention to the socket number and model.
Clock speed supported by the motherboard: The frequency or clock speed of the motherboard determines the number of instructions processed per minute. Here you need to use a closely related clock circuit. Because the CPU clock speed is higher, its power is wasted, and if the CPU clock speed is low, the information processing queue is formed, and there is a possibility of data loss or system slowdown. - RAM transaction speed: RAM has a data transaction speed with processors. This speed is expressed in millions of transactions per second (MT/S). The server CPU you purchase must be fully compatible with this speed. A higher speed means a waste of processor power, and a lower speed means a slowdown of the system or even its failure.
- The number of input channels: Each processor can support a certain number of input channels. In processors, these channels are usually between four and six. It is better to ensure that the selected socket covers all input channels.
Buy a server CPU
When you go to the market or sales sites like ITbaz to buy a server CPU, you will encounter many phrases and words (often in English). For your shopping convenience, we explain some of the most important terms here:
CPU model number (Processor Number)
This number is a combination of model, generation, and processing speed. For example, the Intel ® Core ™ i9 CPU model number is written as i9-12900HX. We have explained the concept of the model number in the article related to it on the ITBaz blog, and you can refer to that section for this topic.
Number of CPU cores (Total Cores)
Each Core consists of several transistors and management and processing sections. The number of CPU cores is always an even number greater than two. The higher the number of CPU cores, the higher the parallel processing power.
Possibility of thread processing (Total Threads)
We said above that threading technology allows the cores to have twice their processing power. If the CPU of the server you buy has such a possibility, this value will be shown as Threads. In front of it, either the option Yes or Enable is registered, meaning it is active and available, or a number is mentioned that indicates the final Power of the processor. This number is often twice the number of CPU cores.
Frequency Power (Base Frequency)
Processing power refers to the number of instructions processed in one CPU second. In today’s CPUs, this value is mentioned in gigahertz (GHz), which refers to billions of instructions processed in one second. Most of the current processors provide you with processing power above 2GHz. Their processing power is equal to or more than two billion command processing per second.
Turbo frequency power
In some server CPUs, a secondary frequency speed called “Turbo Frequency” is mentioned. This frequency is accessible when fast or turbo processor mode can be activated. In this situation, the Power of your CPU will increase drastically and give you Power equal to or less than twice the base frequency of the CPU. This is not good and causes the processor to wear out quickly or even burn it, and it should only be used for a limited period to support heavy processing operations.
Cache value
The cache is a limited but fast memory in the heart of the processor itself. This memory allows you to receive and send information to the CPU cores quickly. In current processors, three levels of cache called L1, L2, and L3 are used. The cache capacity of level one (L1) is often on the order of several kilobytes, while caches of level three (L3) have a capacity of several megabytes. When a processor’s cache value is announced, it often refers to the capacity of the third-level (L3) cache.
CPU power consumption
The power or electricity consumed in the CPU is important for two reasons. First, this part is constantly working, and it can be said that a large part of the energy consumed in your server is consumed in this part. Second, the higher the amount of electricity consumed, the higher the heat loss and heat produced in it will be. The power consumption of processors is expressed in watts (W). The terms Base Power, Turbo Power, and Assure Power are often shown together. It would be best if you based your decision on Turbo Power to choose the right motherboard for your processor.
Memory support
The RAM memory is the main part you should pay attention to when buying a server CPU. Each CPU can support a certain amount of processors and their defined types. Most of the time, the amount of memory support is stated as follows:
- Max Memory Size: Shows the maximum supported RAM for each memory. The higher this amount, the processor cores can support higher capacity RAMs.
- Memory Types: Each CPU can work with one type of RAM. Now there are DDR4 and DDR5 memory types in the market. However, the processors may also work with DDR3 RAM or other models in addition to these two examples described here. Next to this expression, the value of the base speed of the transaction in terms of (MT/s) is also explained. For example, Up To DDR5 4800, MT/s tells you that the selected CPU can support DDR5 RAM up to a maximum transaction speed of 4800 MT/s.
- Memory Channels: Each RAM uses specific communication channels to communicate with the processor. The higher the number of these channels, the faster the data exchange between the CPU and RAM. This value is often even and less than ten.
- Memory Bandwidth: This number shows the amount of information transferred in one second. The higher this value, the more powerful the CPU can receive or send information from the RAM. This number is expressed in GB/s.
- ECC Memory Supported: ECC technology means “error correction technology.” In most of the transactions between the RAM and the CPU, there is a possibility that the transacted data may be corrupted or damaged. To prevent errors, ECC technology is used in memories and processors. CPUs and RAMs that support this feature are used for financial markets, banks, security services, and other sensitive applications.
CPU installation with expansion slots and RAM installation in personal computers
Graphics support
One of the other important characteristics of server CPUs is the support of graphic processing. Processors that have this capability give you good facilities in graphic processing, media playback, and multimedia activities. The main features mentioned here are:
- Graphics processor name: This name is a long phrase. For example, Intel ® UHD Graphics for 12th Gen Intel ® Processors. Most of these refer to the processor’s generation, architecture, and support.
Maximum frequency (Graphic Max Dynamic Frequency): Like the base frequency in CPUs, the higher this value is, the higher the processing speed you will have. - Graphics Output: This option refers to the type of media output channels. We will explain to you about media output channels in a complete article.
- Max Resolution: The final Power of the graphic output is shown this way. It is natural that the higher the resolution and processing frequency, the higher the quality and the better the image. This value may be specified for HDMI, DP, and eDP standards.
- OpenCL support: This is a command library for managing execution commands in parallel. If your processor has this capability, it can support API and related algorithms very well.
Expansion options
Different types of expansion slots (PCIe Slots) are used in a motherboard. Expansion slots allow the installation of different hardware to the motherboard or server chassis. Here are some specifications for CPU expansion slots, which we have explained below:
- Number of Lenses (DMI Lanes): This value indicates the number of base or exchange lenses in a slot. The higher this value, the better the exchange with connected peripherals.
- Slot Generation (PCI Express Revision): Expansion slots are evolving like other server components. This value shows the generation supported by the slot in the purchased CPU. Be sure to match this number with the supported slots on your motherboard.
General specifications of CPUs
You should pay attention to some special features on the CPU package when buying.
- Supported socket type (Socket Support): This number indicates the socket through which a processor can be connected to the motherboard. Models often combine letters and numbers (for example, FCBGA1964). One of the most important things to note here is matching the motherboard socket with the supported socket on the CPU.
- The number of cooperating CPUs (MAX CPU Configuration) indicates the number of processors that can work together on a server simultaneously. This value can be one, two, four, and eight in CPUs. In this case, your server’s CPU can support the processes with good Power and share it with its counterpart.
- Tolerable temperature (T junction): Tolerable temperature indicates the maximum temperature at which the server CPU can work well. The higher this value is, the more capable your CPU is. Most CPUs can withstand temperatures below 100 degrees Celsius. Pay attention that the CPU temperature should not exceed this value and always be at a temperature equal to three-quarters of this number to reach the best possible level.
- CPU Size (Package Size): This value shows the length and width and sometimes the height of the server CPU in millimeters.
In this article, we got acquainted with CPU, CPU specifications, server CPU features, how CPU works, and specifications for purchasing a server CPU. In the next articles in Ibiza, we will fully review all types of specifications, features, technologies, and tips that you should pay attention to when buying a server CPU. We will also provide you with exciting news about all types of CPUs.
common questions
What is a CPU?
CPU or processor (Central Processing Unit) is the main data processor in computers and servers. After processing, the CPU receives different instructions and provides the result to other hardware.
What is a transistor?
The transistor is a current control part that controls electricity flows based on logical commands. In server and computer CPUs, transistors are produced using silicon and small dimensions (between 5 and 14 microns). We have billions of tiny transistors in each CPU that process logical information.
What is the recipe?
Instruction Set is a set of content and logical instructions given to a processor, and the CPU uses it to process information. The unit of instruction is “Word.” Each word can be between 8 and 526 bits long. Instructions can be applied to simple logical, media, and artificial intelligence processing.
What is a CPU core?
The Core is a set of transistors and data flow management parts placed in the main processor. Usually, each processor has two or more cores. The CPU core provides us with the possibility of parallel processing and faster processing. The number of CPU cores is always expressed as an even number of more than two. The higher the number of cores, the higher the processing power of the CPU.
What is parallel processing?
Parallel or Simultaneous Processing means processing several sets of instructions simultaneously in several cores of a CPU. Parallel processes provide us with a higher processing speed and allow us to work with several software at the same time.
What does thread processing mean?
Threaded processing or Hyper-threading means using processing algorithms that increase the Power of CPU cores. This will increase the Power of CPU cores up to two times.
What is CPU cache?
Cache refers to the internal memories of CPUs that enable quick retrieval of information and results from cores. The internal caches of the CPU allow us to increase the speed of processors tremendously due to their proximity and high speed. These memories are faster than RAMs or permanent storage memories such as hard drives and SSDs. The capacity of CPU caches is often between a few megabytes (less than 200 megabytes).
What is CPU graphics?
A Graphics processing unit is a graphics processing unit inside a processor. Due to its architecture and processing power, this unit can facilitate media, image rendering, and support graphic activities. The higher the power of this section, the better the processor can support graphics media.
What is a motherboard socket?
The motherboard socket is where the CPU is connected to the motherboard. Each processor is designed based on a specific socket and can only be used on the desired socket. Sockets are a place to hold the body and pins to exchange Power and data.
What is CPU frequency?
CPU Base Frequency refers to the processing speed of instructions in processors in one second. The higher the number of commands processed in one second, the faster the processing speed. The basis of this value is GHz or gigahertz, which means billions of instructions per second. Current processors have a processing power of 2GHz or more.