Some CPU concepts are introduced.

Source: Internet
Author: User
The full name of CPU is central processing unit, that is, the central processor. From the initial appearance to the growth of the CPU, as the manufacturing technology is becoming more and more advanced, more and more electronic components are integrated in it, from tens of thousands to now, even tens of millions of micro transistors constitute the internal structure of the CPU. So how does tens of millions of transistors work?

The internal structure of the CPU can be divided into three parts: Control Unit, logic unit, and memory unit. The working principle of CPU is similar to the processing process of products in a factory: from the raw materials (Instructions) that enter the factory to the scheduling and distribution of material distribution departments (control units, it is sent to the production line (logical Computing Unit), produced (processed data), and then stored in the warehouse (storage, finally, wait for the product to be sold on the market (for use by applications ). As the core of the entire microcomputer system, CPU is often synonymous with computers of various grades, such as 286, 386, 486, 586/Pentium, to today's PentiumII, K6-2/3, cyrix II, pentiumiii, althon (K7), celeronii, duron, via cyrixiii and the upcoming intel pentiumiv, etc, CPU is an important indicator of the performance of the computer it is configured with. Here we will introduce some CPU-related performance indicators:

Clock speed (main CLK), frequency doubling, and external frequency. I often hear people say, "What is the CPU frequency ......" In fact, the frequency refers to the CPU clock speed, that is, the CPU clock frequency. The full name is CPU clock speed, that is, the operating frequency during CPU operation. Generally, the higher the clock speed, the more commands completed in a clock cycle, and the faster the CPU speed. However, because various CPUs have different internal structures, not all CPUs with the same clock frequency have the same performance. The external frequency is the operating frequency of the system bus, while the frequency doubling is the multiple of the difference between the external frequency of the CPU and the main frequency. The relationship is very close: frequency = frequency X frequency doubling. In general, the super frequency is the Super frequency, that is, to make the CPU work at a higher clock speed. Generally, the super frequency can start from two aspects: Ultra-frequency and super-frequency, the current CPU has usually locked the frequency doubling, so it is generally an extra-high frequency, which is related to the motherboard. The extra-frequency can be adjusted through the dip switch, or through the soft menu. During overclocking, you often need to slightly increase the operating voltage of the CPU and pay attention to the heat dissipation of the CPU. Overclocking is also closely related to the motherboard, video card, hard disk, memory, power supply, etc. Be sure to avoid burning the CPU.

Memory-bus speed (memory bus speed ). The data processed by the CPU is from the primary memory (memory). Generally, the data stored on the external storage (disk or various storage media) must pass through the memory and then enter the CPU for processing. So the channel between it and the memory ?? The speed of the memory bus is very important to the overall system performance. Because the running speed between the memory and the CPU is more or less different, there is a level-2 cache to coordinate the differences between the two, the memory bus speed refers to the communication speed between the CPU and the L2 high-speed cache (l2cache) and the memory. At present, generally, the desktop CPU is 512 K (half speed)/256 K (full speed) l2cache, while for Intel celon, it is on-die (full speed) 128 K l2cache, AMD althon/K7, K6-3 is known to be attached up to 8 m l2cache.

Expansion-bus speed (Extended bus speed ). Extended bus refers to the local bus installed on the microcomputer system, such as VESA, PCI, or AGP bus. when we open the computer, we will see something like a slot. These are the expansion slots, the extended bus serves as a bridge between the CPU and these external devices.

Supply voltage ). Any electrical appliance requires electricity when it is working, and there will naturally be rated voltage, and the CPU is no exception. The operating voltage refers to the voltage required for normal CPU operation. In the early-era, the operating voltage of the CPU was generally 5 V, because the manufacturing technology at that time was relatively backward, so that the CPU generated too much heat and reduced the service life. With the increase of CPU manufacturing technology and clock speed, the operating voltage of various CPUs has gradually declined in recent years to solve the problem of excessive heat dissipation, at present, the average desktop CPU operating voltage is 3.3 V/2.2 V, while the notebook dedicated CPU has a lower operating voltage, or even 1.2 V, which greatly reduces power consumption, while the heat dissipation is also greatly reduced, of course, its production cost is also greatly increased.

Address Bus width (address bus width ). The address bus width determines the physical address space that the CPU can access. Simply put, it means that the CPU can use a large amount of memory. We don't need to talk about 16-bit microcomputer, but for a microcomputer system above 386, the address line is 32-bit in width, and theoretically a maximum of 4096 MB (4 GB) of physical space can be directly accessed. Of course, this is related to the number of slots on the motherboard and the maximum size of memory supported by each slot.

Data Bus width ). The data bus is responsible for the data traffic of the entire system. The data bus width determines the amount of data transmitted between the CPU and the second-level high-speed cache, memory, and input/output devices.

Co-processor (Digital coprocessor ). There is no built-in coprocessor In the CPU before 486. Because the main function of the coprocessor is to perform floating-point operations, the floating-point operation performance of 386, 286, 8088, and so on is quite poor, I believe that 386 of my friends know that an external coprocessor can be added to the motherboard to enhance the floating point operation function. Since 486, CPU generally has built-in coprocessor, and its functions are no longer limited to enhanced floating-point operations. CPU with built-in coprocessor can accelerate the computing of specific types of values, some software systems that require complex computing, such as the later version of Auto CAD, require coprocessor support.

Super Sclar exceeds the standard. Exceeding the threshold means that the CPU can execute more than one instruction within a clock cycle. This is hard to imagine on 486 or the previous CPU. Only the CPU above the Pentium level has the structure of this amount exceeding the standard; the CPU below 486 belongs to the low scale structure, that is to say, executing an instruction in such a cpu requires at least one or more clock cycles.

L1cache, which we often call a level-1 high-speed cache. The built-in high-speed cache in the CPU can improve the CPU running efficiency, which is precisely why 486dlc is faster than mongodx-40. The built-in L1 cache capacity and structure have a greater impact on CPU performance. The larger the capacity, the higher the performance, so this is why some companies strive to increase L1-level high-speed buffer storage capacity. However, high-speed buffer memory is composed of static RAM and has a complex structure. When the CPU core area cannot be too large, the l1-level high-speed cache capacity cannot be too large. L1cache is generally 32 KB, but the l1cache capacity has been increased in AMD's althon/K7 series.

Write back (with a write-back structure. It is effective for read and write operations and is fast. The write-through structure of the cache is only valid for read operations.

Dynamic procedure (Dynamic processing ). Dynamic processing is a new technology applied in high-energy Pentium processors. It creatively integrates three technologies designed to improve the processing efficiency of data operations. These three technologies are multi-channel Traffic Distribution Prediction, data traffic analysis, and prediction execution. Dynamic processing does not simply execute a string of commands, but improves the processing efficiency by operating data. Dynamic processing includes multi-channel Traffic Distribution Prediction: the flow direction of the program is predicted using several branches. After the multi-channel traffic distribution prediction algorithm is adopted, the processor can jump to the command flow. It predicts that the accuracy of the next instruction in the memory can reach an astonishing 90% or more. This is because when the processor obtains commands, it will also look for instructions to be executed in the future in the program. This technology accelerates Task Transfer to the processor. 2. Data Traffic Analysis: Analyze and rearrange the commands based on the original program sequence to optimize the execution sequence: The processor reads decoded software commands, determine whether the command can be processed or whether it must be processed together with other commands. Then, the processor determines how to optimize the execution sequence for efficient processing and execution of commands. 3. Speculative execution: Improves the execution speed by interpreting and executing program commands that are possibly required in advance: When the processor executes the commands (five at a time ), the method of "speculative execution" is used. In this way, the super processing capability of the Pentium II processor can be fully utilized to improve the software performance. The software instruction to be processed is based on the speculative branch, so the result is retained as the prediction result. Once its final state can be determined, the command can return to its normal order and maintain its permanent state.

Contact Us

The content source of this page is from Internet, which doesn't represent Alibaba Cloud's opinion; products and services mentioned on that page don't have any relationship with Alibaba Cloud. If the content of the page makes you feel confusing, please write us an email, we will handle the problem within 5 days after receiving your email.

If you find any instances of plagiarism from the community, please send an email to: and provide relevant evidence. A staff member will contact you within 5 working days.

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.