What is the number of processor kernels in the phone, for which it is responsible, what function does it perform? What does the number of nuclei in the smartphone affect? What is the largest number of nuclei in the smartphone? What does the number of kernels in the phone, smartphone? Number of Cores .

I told why the growth of processor frequency stalled on several gigahertians. Now let's talk about why the development of the number of nuclei in user processors is also extremely slow: so, the first honest dual-core processor (where both cores were in one crystal), built on the X86 architecture, appeared as much as in 2006, 12 years ago There was a ruler Intel Core Duo. And since then 2- nuclear processors From the arena do not go, moreover - actively develop: so, just the other day it came out lenovo laptop With the processor built on the newest (for architecture x86) 10 nm technical process. And yes, as you already guessed, this processor has exactly 2 kernels.

For user processors, the number of nuclei stalled for 6 from 2010, with the output of the AMD Phenom X6 line - yes, AMD FX was not honest 8-nuclear processors (there were 4 APUs), as well as Ryzen 7 is two blocks of 4 kernels located side by side on the crystal. And here, of course, the question arises - why so? After all, the same video cards, being in 1995-6, in fact, "single-headed" (that is, those who had 1 shader), managed to increase their number to several thousand to the current time - so, in NVIDIA Titan V them as much as 5120! At the same time, for a much larger term of development of the X86 architecture, user processors stopped on honest 6 cores on a crystal, and CPU for high-performance PCs - by 18, that is, a couple of orders of magnitude less than that of video cards. Why? About this and talk below.

Architecture CPU.

Initially, everything intel processors The X86 was built on the CISC architecture (Complex Instructions Set Computing, processors with a full set of instructions) - that is, they are implemented in them the maximum number of instructions "for all occasions". On the one hand, it's great: so, in the 90s CPU answered and for rendering pictures, and even for the sound (there was such a lifehak - if the game slows down, then the sound in it can help). And even now, the processor is a form of a combine that can all - and the same is a problem: to paralle the random task for several nuclei - the task is not trivial. Suppose you can simply be done with two nuclei: one core "hang" the system and all background tasks, to another - only the application. It will always work, but the performance increase will be far from double, since usually background processes require significantly less resources than the current hard task.

Left - GPU scheme NVIDIA GTX. 980 Ti, where you can see 2816 CUDA-nuclei, combined into clusters. On the right - the photography of the AMD Ryzen processor crystal, where 4 large kernels can be seen.

And now we will imagine that we have not two, but 4 or in general 8 cores. Yes, in archiving tasks and other calculations, parallelization works well (and that is why the same server processors can have several dozen nuclei). But what if we have a challenge with a random outcome (which, alas, most) - let's say a game? After all, each new action depends entirely on the player, so the "scattering" of such a load on several nuclei - the task is not simple, because of which the developers are often prescribed by the "hands" than the kernels are engaged in: so, for example, one can only be occupied only Processing action artificial IntelligenceOther respond only for surround sound, and so on. To load even the 8-nuclear processor in this way - it is almost impossible that we are visible in practice.

With video cards, it is also simpler: GPU, in fact, engaged in calculations and only by them, and the number of calculations is limited and small. Therefore, firstly, you can optimize the computing kernels themselves (NVIDIA they are called CUDA) it is under the necessary tasks, and, secondly, all possible tasks are known, then the process of their parallelization of difficulties does not cause difficulties. And thirdly, control is not separate shaiders, but by computing modules, which include 64-192 shader, so big number Shaders are not a problem.

Energy consumption

One of the reasons for the abandon from the following frequency racing is a sharp increase in power consumption. As I have already explained in the article with a slowdown in the growth of the CPU frequency, the heat generation of the processor is proportional to the frequency cube. In other words, if at a frequency of 2 GHz, the processor allocates 100 W heat, which, in principle, you can easily remove with an air cooler, then 4 GHz will work already 800 W, which is possible to take off at best, the evaporative chamber with liquid nitrogen (although it should be considered that the formula is still approximate, and in the processor there are not only computing kernels, but it is possible to obtain the order of numbers with its help).

Therefore, the growth of stirre was a great way out: so, roughly speaking, two-core 2 GHz processor will consume 200 W, but the one-core 3 GHz is almost 340, that is, the winner of the heat generation is more than 50%, while in tasks with good optimization under multithreading Low-frequency dual-core CPU will still be faster than high-frequency single-core.

An example of an evaporative chamber with liquid nitrogen for cooling extremely overclocked CPUs.

It would seem - this is a golden bottom, quickly make a 10-core processor with a frequency of 1 GHz, which will allocate only 25% more heat than one-core CPU with 2 GHz (if 2 GHz processor allocates 100 W heat, then 1 GHz - Total 12.5 W, 10 nuclei - about 125 W). But here we quickly rest in the fact that not all tasks are well paralleled, so in practice it will often be obtained so that a much cheaper one-core CPU CPU with 2 GHz will be significantly faster than a much more expensive 10-nuclear one, but from 1 GHz. But still there are such processors - in the server segment, where there are no problems with parallelization tasks, and 40-60 nuclear CPUs with 1.5 GHz frequencies often turn out to be at times faster than 8-10 nuclear processors with frequencies under 4 GHz, highlighting a comparable number Heat.

Therefore, CPU manufacturers have to ensure that the nuclei does not suffer to single-threaded performance, and taking into account the fact that the heat removal limit in the usual home PC was "Nashchupan" for a long time (this is about 60-100 W) - ways to increase the number of cores With the same single-core performance and the same heat release only two: this is or optimized by the processor architecture itself, increasing its performance for the tact, or reduce the technical process. But, alas, and the other goes all slower: For more than 30 years of the existence of X86, the processors "polished" are already almost all that can, so the increase is in the best case 5% per generation, and the reduction in the process is increasingly difficult due to Fundamental problems of creating correctly functioning transistors (quantum effects are already beginning to affect the size of a dozen nanometers, it is difficult to produce a suitable laser, etc.) - therefore, alas, to increase the number of cores is increasingly more difficult.

Crystal size

If we look at the processor crystals area 15 years ago, we will see that it is only about 100-150 square millimeters. About 5-7 years ago, Dorosli chips up to 300-400 sq. MM and ... the process almost stopped. Why? Everything is simple - first, it is very difficult to produce gigantic crystals, which is why the amount of marriage increases sharply, and, it means that the final cost of the CPU.

Secondly, the fragility increases: a large crystal can very easy to split, besides, its different edges can be sink differently, which is why his physical damage can again occur.

Comparison of crystals Intel Pentium. 3 and Core i9.

Well, thirdly - the speed of light also contributes to the restriction: yes, it is although it is not infinite, and it can make a delay with large crystals, but it is impossible to do the work of the processor.

Eventually maximum size The crystal stopped somewhere on 500 sq. MM, and it is hardly to grow - therefore to increase the number of nuclei, it is necessary to reduce their size. It would seem - the same NVIDIA or AMD could do it, and their GPUs have thousands of shaders. But here it should be understood that the shaders are not full nuclei - for example, they do not have their own cache, but only the overall, plus the "sharpening" under certain tasks allowed "throw out" of them all too much that again it affected their size. And the CPU does not only have a full-fledged kernel with their own cache, but often on the same crystal there is both graphics, and various controllers - so that in the end, again, almost single ways Increasing the number of nuclei with the same crystal size - this is all the same optimization and all the same decrease in the process, and they, as I have already written, go slowly.

Optimization of work

Imagine that we have a team of people performing various tasks, some of which require the work of several people at the same time. If people are in it two - they will be able to negotiate and work effectively. Four is more complicated, but the work will also be quite effective. And if people are 10, or even 20? There is already some kind of means of communication between them, otherwise the "Dissolves" will be found in the work when someone will be not busy anything. In the processors from Intel, this means of communication is the ring bus that binds all the kernels and allows them to exchange information among themselves.

But even it does not help: so, with the same frequencies, the 10-nuclear and 18-core processors from Intel generation Skylake-x differ in terms of performance by only 25-30%, although they must increase by 80% in theory. The reason is just in the bus - no matter how good it is, there will still be delays and downtime, and the more nuclei - the worse there will be a situation. But why then there are no such problems in video cards? Everything is simple - if the processor core can be submitted by people who can perform various tasks, then the video card computing blocks are rather robots on the conveyor that can only be executed. They are essentially "to negotiate" - therefore, with the increase in their number, the effectiveness drops slower: so, the difference in CUDA between 1080 (2560 pieces) and 1080 Ti (3584 pieces) - 40%, in practice about 25-35%, then There are losses significantly less.

The more nuclei, the worse they work together, right up to zero performance increases with increasing number of cores.

Therefore, the number of special meaning nuclei to increase no - the increase from each new kernel will be all lower. Moreover, it is difficult to solve this problem - it is necessary to develop a tire that would allow to transmit data between any two nuclei with the same delay. In this case, the topology of the star is suitable in this case - when all the kernels must be connected to the hub, but no one has done such a realization yet.

So, as a result, as we see that the incidence of the frequency is that the increase in the number of nuclei - the task is quite complicated, and the game often does not have a candle. And in the near future, it is unlikely that something will be seriously changed, since nothing better than silicon crystals have not yet invented.

But with the conquest of new vertices of frequency indicators, it became harder to increase it, since it affected the increase in TDP processors. Therefore, the developers began to raise processors in width, namely add the nuclei and the concept of multi-core appeared.

More literally 6-7 years ago, the processor multi-daidity was practically not heard. No, multi-core processors from the same IBM company existed earlier, but the appearance of the first dual-core processor for desktop computers It took place only in 2005, and the Pentium D processor was called. Also, in 2005, the Opteron Opteron dial was released, but for server systems.

In this article, we will not delve detail into historical facts, and we will discuss modern multi-core processors as one of the characteristics of the CPU. And most importantly - we need to deal with what this multi-core in terms of performance for the processor and for us with you.

Increased productivity at the expense of multi-core

The principle of increasing the performance of the processor at the expense of several cores is to deal with the execution of streams (various tasks) into several cores. Summarizing, it can be said that almost every process running in your system has several streams.

Immediately make a reservation that the operating system can virtually create many streams for itself and perform it all as it would be at the same time, even if physically processor and one-core. This principle implements that the most multitasking of Windows (for example, simultaneous listening to music and text set).

Take for example an antivirus program. One stream we will have a computer scanning, the other - update antivirus base (We are all very simplified, in order to understand the overall concept).

And consider what will be in two different cases:

a) single-core processor. Since two streams are performed with us at the same time, you need to create this very simultaneous execution for the user (visually). Operating system, makes a cider:switching between the execution of these two streams is occurs (these switching instantaneous and time goes in milliseconds). That is, the system slightly "boost" update, then sharply switched to scan, then back to the update. Thus, for us, you have the impression of the simultaneous execution of these two tasks. But what is lost? Of course, performance. So let's consider the second option.

b) multi-core processor. In this case, this switch will not be. The system will clearly send each stream to a separate kernel, which will increase us to get rid of the destructive switching of the flow from flow to flow (idealizing the situation). Two streams are performed simultaneously, this is the principle of multi-purpose and multithreading. Ultimately, we will make a lot of scanning and update on a multi-core processor than one-core. But there is a harness - not all programs support multi-core. Not every program can be optimized in this way. And everything happens far away, as far as we described. But every day, developers create more and more programs that have a perfectly optimized code, to perform on multi-core processors.

Do you need multi-core processors? Casual response

For choice of processor For a computer (namely, when reflected on the number of cores), it is necessary to determine the main types of tasks that it will perform.

To improve knowledge in the field computer Iron., you can get acquainted with the material about processor sockets .

The point of the start can be called dual-core processors, as it makes no sense to return to single-core solutions. But dual-core processors are different. This may not be the "most" fresh Celeron, and may be Core i3 on Ivy Bridge, just like AMD - SEMPRON or Phenom II. Naturally, due to other indicators, their performance will be very different, so you need to look at everything comprehensively and compared the multi-core with others characteristics of processors.

For example, at Core i3 on Ivy Bridge, there is a Hyper-Treading technology, which allows the 4 stream to handle simultaneously (the operating system sees 4 logical kernels, instead of 2 physical). And the same Celeron does not boast so much.

But back directly to reflections regarding the required tasks. If the computer is necessary for office work and surfing on the Internet, then the dual-core processor is enough with his head.

When it comes to gaming performance, here is to comfortably feel in most games you need 4 nuclei and more. But then the same picking pops: not all the games have an optimized code under 4 nuclear processors, and if optimized, then not so effectively, as if I wanted. But, in principle, for the games now the optimal solution is exactly the 4th nuclear processor.

Today, the same 8 nuclear aMD processors For games, it is redundant, it is excessively the number of cores, but the performance does not reach, but they have other advantages. These are the most 8 cores, very much help in tasks where necessary powerful work with high-quality multi-threaded load. This can be attributed, for example, rendering (miscalculation) video, or server calculations. Therefore, for such tasks is required 6, 8 or more nuclei. Yes, and soon the game will be able to load 8 and more nuclei qualitatively, so in the future, everything is very rosy.

Do not forget that there are a lot of tasks creating a single-threaded load. And you should ask yourself a question: I need this 8 nadher or not?

Summing up the small results, I will once again note that the benefits of multi-core are manifested with "weighty" computational multi-threaded work. And if you do not play games with the prospecting requirements and do not engage in specific types of work requiring good computing power, then spend money on expensive multi-core processors, it simply makes no sense (

QX | July 22, 2015, 14:45
Not only frequency, technical process too. Modern 2-core processors of 3 GHz do not compare with the first 2-nuclear, from those that are also 3 GHz. The frequency is the same, but old just creepy brakes in comparison with new ones. As a result, the modern 2-core i3 is much better than the 4-kernel Quad Q6600. Even Pentium G is better than an old quad.

QX | July 11, 2015, 12:18
Here the frequency difference is not large, 3.5 against 3 GHz. Therefore, 4 nuclei are interesting. But of course, if the remaining characteristics also do not lag behind. Many nuclei need to archive, video encoding, etc. Taking 2 kernels also save, slightly. Another question how much will work on it. Well, it would be better if both models were specifically called. And so, I would advise you Core i3 more powerful and fresher.

Makos007 | March 30, 2015, 16:00
I'm going to grow out on the tree. Therefore, I will immediately say - your choice of a dual-core processor with a higher frequency. If the theory is not interesting, then you can not read further.

The processor frequency is, in fact, the number of operations performed by it per unit of time. Thus, the higher the frequency, the more actions are performed per second, for example.

As for us with the number of kernels ... if there is more than one kernel, the processor can determine more than one task. It is like conveyor ribbons. One conveyor tape works quickly, but two parallel tapes on which the operations are coming out twice as a result. So in the theory, dual-core solutions will operate twice as fast.

This is a theory, but as with the conveyors, these two streams need to load something. At the same time loading correctly so that each tape worked with complete return. In the case of processors, it depends on the architecture of programs and games that use this very multi-core. If the application knows how to separate tasks for several threads (read - use multi-core processor), then multi-core can give a significant increase in the velocity of commands. And if it does not know how to task such that it is impossible to divide, then there is a lot of nuclei in CPU completely or not.

In fact, the question of the optimal number of cores is complex. The architecture of the cores themselves and connections between them is still important. So the first multi-core processors had significantly less functional devicethan modern. In addition, it should be borne in mind that the modern Windows 7 OS and Windows 8 (I do not consider here * NIX systems and their support for multi-core processors - a separate and very interesting topic) exit very well to parallelize many tasks. Thus, multi-core helps not slow down the main processes (used by the user applications and games) due to the performance of background tasks. Thus, antivirus protection and firewall will not slow down (more precisely, to a lesser extent will slow down) the launched game or work in photoshop.

For which programs are important multi-trial. After spending some time on the Internet, it can be found out that it speeds up conversion of video and audio; Renéring 3D models, signal encryption, etc. You do not need 4 kernels to work in Photoshop and video editing. It is enough, as I said, two, but with higher speed of each of them.

teleport. | April 21, 2013, 01:30
A simple performance calculation shows: for 2 nuclear total performance 2 x 3.5 \u003d 7, for 4 nuclear - 4 x 3 \u003d 12. So the 4-core almost 2 times more powerful. In addition, he is probably modern, which means more economical and more efficient. And if only one kernel is used - it is heated less, since the frequency of one nucleus is slightly lower, but it is essential for heating.

For a video editing, the processor is most likely not critical there is mainly the resources of the video card or a special video editing board. But the processor in this also participates and if the 2nd nuclear will give one nucleus under this task, then the rest of the tasks (different antiviruses) will fight for the remaining kernel, which will lead to terrible stupidity. In short, multi-core is better.

yang. | April 11, 2013, 20:22
In this case, the dual-core processor will be more efficient and more economical in all respects.

Instruction

If you have installed operating windows system, Find out how much nuclei in your processor can be through properties. To do this, select the Computer icon on the desktop, press Alt + Enter or the right mouse button and in context menu "Properties".

Window with information about operating system, processor, random access memory and computer name. On the right there will be references, among which you need to find "Device Manager".

The dispatcher will indicate the equipment that you have installed. Find the Processor item in the list and click on the arrow next to it. A column unfolds, in which the number of your processors will be indicated.

You can run the task manager using Ctrl + Shift + ESC combination. Open the tab called "Speed". The number of windows in the section "TsP loading chronology" corresponds to the number of cores of your processor.

If the computer is enabled on the computer multi-core processor, then the task manager will show the number of symotic cores. This can be defined if all the kernels show a completely identical load. Then you can use free cPU-Z utility. The CPU tab shows all information about the processor. Below is the Core window where the number of cores is indicated.

You can use one more free program PC Wizard. It can be downloaded from the developer's website. Install the program to the computer. Run the PC Wizard.exe file, click the "Iron" tab, then the "processor". Find the "Element" section on the right, and in it the Number of Core. The "Description" section shows the number of cores.

Short story on simple language On the kernels of mobile processors, their functions and the required quantity.

Navigation

For any person who is decided to acquire a new smartphone, the main criterion of choice is not only the price, but also the power of the gadget. If you go to the site of some online store and open the technical characteristics of smartphones, then among them you can see such a definition as "processor".

Many, even technically illiterate users, this detail is familiar and they have an idea of \u200b\u200bwhat function it performs. However, standing next to him the words "dual-core" or "quad-core" cause many bewilders.

In our article we will talk about what the core of the processor in the smartphone is for which it replies and truthfully the opinion is that the more nuclei in the processor, the more powerful telephone.

What is a processor in the phone?

Before switching to the nuclei, first need to understand what the processor is. The processor is a miniature device that is responsible for mathematical, logical and managerial operations made by man in the machine code.
As a rule, the processor is made in the form of a single integrated circuit, the basis of which is a silicon chip and a huge number located on it, transistors. In some cases, the processor can consist of two or more specialized chips.

The speed or the power of the processor directly depends on total Transistors deposited on a silicon chip. The power of the processor is measured in the clock frequency ( GHz.) And the more transistors are applied on the silicon chip, the higher the processor (power) clock frequency will be.
However, which comes in transistors of the current, has a property to heat a silicon chip, which under the influence of high temperatures fails. And the more transistors are located on the chip, the faster it heats up and reaches its thermal limit. Just in order to avoid overheating, processors with two or more nuclei were invented.

What is the processor core in the smartphone and what are they responsible for?

The kernel is the main processor module where all the information is processed and calculations are made. If you draw an analogy with the human body, the processor is the brain, and the nuclei is its hemispheres. The human brain has two, but the number of smartphone processor cores can reach eight pieces.

Above, we talked about the fact that the power of the processor depends on the number of transistors applied to it and mentioned overheating. The presence of several cores in the processor is necessary in order to distribute the load on the processor between them and reduce heat transfer.
Thus, if one kernel does not cope with the flow of information being processed, the second core is automatically activated and a part of the work will take place, thereby preventing overheating. The presence of two or more nuclei in the processor allows you to apply more transistors on it and accordingly increase its power or data processing speed.

What does the number of nuclei in the smartphone affect?

As we have already found out, the kernels help unload the processor, reduce heat transfer and increase its speed. Thus, the more in the core processor installed on your phone, the more actions you can perform simultaneously.

For example, if you having a smartphone with a single-core processor, play it into the game and want to start some second application in parallel, then your game will be automatically closed, since the processor cannot simultaneously process such big flow data.
If you do the same on a dual-core processor, then one of his kernel will take on the job of the game, and the second will process the running application.
There are also heavy applications that are downloaded by several processor cores at once. They are called multi-threaded. These include heavy games and some graphic editor. If you try to run such an application on your smartphone with a single-core processor, then at best it will simply not start. The worst layout can be full of breathing and overheating of the device.

What is the largest number of nuclei in the smartphone?

Today in cell phones and tablets are installed processors with maximum number Nuclei of ten pieces. Surely they could be more, but the developers do not see the need for this time.
But, despite the point of view of processor manufacturers, many analysts and experts adhere to the opinions that the future of gadgets is behind their multitasking, which is impossible without the presence of multi-core processors.

How many kernels in the phone, the smartphone is better?

Many buyers adhere to the opinions that the eight-core processor is two times more powerful than quad-core. If we consider it from the point of view of logic and not go into the details of the processor device, then eight more than four, and therefore the power of the gadget will be higher. However, this opinion is in the root of erroneous.
As mentioned, the number of processor cores increases the speed of the smartphone due to the uniform distribution of processes performed simultaneously. But most of the existing today mobile applications They are single threaded and at the same time can only use the processor core. In rare cases, two.

Multi-core processors are needed only if you play heavy games that give high load on the processor and are able to use four nuclei and more. Such games today are units, since the developers of the gaming industry are trying to optimize their products even under weak devices in order to increase sales.
You can not give a clear answer to the question in the title. It all depends on your needs and technical characteristics Devices in general. If you need good smartphone For games, it is worth paying attention not only to the number of processor cores, but also on its clock frequency, as well as the amount of RAM.

For example, a smartphone with 4 GB RAM, four-core processor and clock frequency 1.7 GHz It will be much faster than a similar smartphone with an eight-core processor and clock frequency 1 GHz.
The processor device also plays an important role. Each manufacturer has a processor structure in different ways. For example, processors from manufacturers Atom and Snapdragon. With the same number of cores and clock frequencies will be different with each other.