What is a server CPU?

The server CPU is the CPU (central processing unit) used on the server. Server is an important device in the network. It must accept access from tens to thousands of people, therefore it has strict requirements for fast throughput, ultra stability, and long-term operation of large amounts of data. Therefore, the CPU is the "brain" of a computer and the main indicator of server performance.

1、 Classification of server CPUs

The CPU of the server is still distinguished by the instruction system of the CPU, which is generally divided into two types: CISC CPU and RISC CPU. Later, there was a 64 bit VLIW (Very long instruction word) instruction system CPU.

CISC server C

CISC is the abbreviation for "Complex Instruction Set Computing". It refers to the x86 (Intel CPU Naming Specification) series of CPUs produced by Intel and their compatible CPUs (CISC), based on PC (Personal Computer) architecture. CISC type CPUs are mainly Intel server CPUs and AMD server CPUs.

Intel Server CPU Xeon

advantage

From the current development status of servers, IA architecture (CISC architecture) PC servers are characterized by being "small, compact, and stable". Due to their reliable performance and low price, they have been widely used. In the fields of the Internet and local area networks, they are used for file services, printing services, communication services, web services, email services, database services, application services, etc.

shortcoming

The biggest drawback of IA-64 microprocessors is their lack of compatibility with x86. In order to make IA-64 processors run two generations of software better, Intel has introduced an x86-to-IA-64 decoder on IA-64 (Itanium, Itanium2...) processors to convert x86 instructions into IA-64 instructions. This decoder is not the most efficient decoder, nor is it the best way to run x86 code (the best way is to run x86 code directly on the x86 processor), so Itanium and Itanium2 perform very poorly when running x86 applications. This became the root cause of the generation of X86-64.

Finally, it is worth noting that although the CPU is one of the most important factors determining server performance, it cannot perform as expected without the support and coordination of other accessories.

The biggest drawback of the IA-64 microprocessor is its lack of compatibility with x86. On the other hand, Intel has introduced x86 to IA-64 decoders on IA-64 processors such as Itanium and Itanium2 to enable IA-64 chips to better run software from these two eras. This enables the conversion of x86 instructions into IA-64 instructions. This decoder is not the most efficient decoder, nor is it the best way to run x86 code (the best way is to run x86 code directly on the x86 processor), so Itanium and Itanium2 perform very poorly when running x86 applications. This became the root cause of the generation of X86-64.

Although CPU is one of the most important factors determining server performance, without the support and cooperation of other accessories, it cannot perform as expected.

RISC server CPU

RISC is the abbreviation for "Reduced Instruction Set Computing". It was developed on the basis of the CISC (Complex Instruction Set Computer) instruction system, and some people tested the CISC machine, indicating that the frequency of various instructions is completely different. The most commonly used are some relatively simple instructions, which only account for 20% of the total number of instructions, but have a frequency of 80% in the program. A complex instruction system will inevitably increase the complexity of microprocessors, resulting in long development time and high costs. And complex instructions require complex operations, which will inevitably reduce the speed of the computer.

Due to these reasons, RISC CPUs were born in the 1980s. Compared to CISC CPUs, they not only simplify the instruction system, but also adopt a structure called "superscalar and hyperpipelined", greatly improving parallel processing capabilities. (Parallel processing refers to a server having multiple CPUs processing simultaneously. Parallel processing can greatly improve the server's data processing ability. Departmental and enterprise level servers should support CPU parallel processing technology.) In other words, this architecture has much higher performance than CPUs with CISC architecture at the same frequency, which is determined by the technical characteristics of CPUs. The RISC instruction system is more suitable for the operating system of advanced servers, UNIX, and Linux is also an operating system similar to UNIX.

The main types of CPUs that use RISC instructions in mid to advanced servers are as follows.

(1) PowerPC processor

(2) SPARC processor

(3) PA-RISC processor

(4) MIPS processo

(5) Alpha processor

Currently, the two main CPU processor instruction systems, CISC and RI

The architecture is leveraging each other's strengths and weaknesses and developing towards convergence. CISC borrowed the ideas of RISC to optimize the efficiency of instruction systems, while RISC introduced enhanced instructions to improve the efficiency of complex task processing. Therefore, there is no need to overly focus on the differences between CISC and RISC, as both architectures are highly advanced and will evolve over time.

The characteristic of CISC complex instruction set is the large number of instructions, where one instruction executes multiple functions. Its advantage is the high efficiency in performing specific functions such as multimedia processing. The disadvantage is that the system design is complex and the execution efficiency is low. Typical architectures include x86.

The characteristic of RISC reduced instruction set is that there are fewer instructions, and complex tasks are completed by combining multiple reduced instructions. Its advantages are high execution efficiency and low power consumption for ordinary tasks. The disadvantage is that for some complex tasks, such as multimedia processing, the processing efficiency is low. Typical architectures include ARM, Power, MIPS, Alpha, and SPARC.

VLIW server CPU

VLIW is the abbreviation of "Very long instruction word". The VLIW architecture adopts advanced EPIC (Explicit Parallel Instruction) design, which we also refer to as the "IA-64 architecture". The EPIC processors are mainly Intel's IA-64 (including Intel's Anthem processor) and AMD's x86-64.

II Server CPU and Desktop CPU

1. Different instruction sets

The common CPUs equipped with home or work computers are usually CISC complex instruction sets, which pursue large and comprehensive instruction sets and attempt to integrate various common functions as a whole. However, compared to server CPUs, their call speed and hit rate are lower. The instructions for the server CPU are usually RISC (Reduced Instruction Set). The advantage of this design is that it is more targeted, can be optimized for different needs, and is more energy-efficient.

2. Different Caches

Caching also determines the performance of the CPU. Due to the high computational performance requirements of server CPUs, they typically apply the most advanced processes and technologies and are equipped with one, two, or three caches for more powerful operation. The server CPU has been using the third level cache for a long time. In recent years, cache technology has only been used by ordinary CPUs.

3. Different interfaces

The server CPU and regular CPU interfaces are usually different. Compared with ordinary CPU interfaces, the server CPU interfaces mainly include Socket 771, Socket 775, LGA 2011, and LGA 1150. The motherboard equipped with a server CPU usually does not have a graphics card slot, as the CPU is equipped with a core graphics card to meet the requirements, and its CPU bus bandwidth is higher than that of a household CPU.

4. Different stability requirements

The server CPU is designed to work stably for a long time and is basically designed to work continuously throughout the year. Server CPUs differ greatly from household CPUs in terms of stability and reliability. Servers typically run 365 days a year with occasional downtime for maintenance, which requires high stability.

A regular CPU is designed to work continuously for 72 hours.

5. Multiple interconnection supports different

Multiplexing is a technology used for servers. For example, the server motherboard can have multiple CPU socket at the same time, allowing multiple CPUs to be installed at the same time. This is the CPU multi-channel interconnection technology. This technology is only supported by the server CPU, while ordinary home computers (with only one CPU installed on the motherboard) do not support multiplexing.

6. Different Prices

Multiplexing is a technology used for servers. For example, the server motherboard can have multiple CPU socket at the same time, allowing multiple CPUs to be installed at the same time. This is the CPU multi-channel interconnection technology. This technology is only supported by the server CPU, while ordinary home computers (with only one CPU installed on the motherboard) do not support multiplexing.

3、 Top 10 Popular Server CPUs

1 Intel Xeon E5-2678 v3

2 AMD Skyline 7763

3 Intel Xeon E5-2680 v2

4 Intel Zhiqiang Silver 4210R

5 Intel Xeon Gold Medal 5218

6 Intel Xeon E5-2650 v2

7 Intel Xeon E5-2680 v2

8 Intel Zhiqiang Silver 4210

9 Intel Xeon Gold 5218R

10 Intel Xeon E5-2670 v2