Main characteristics of the central processor

The processor is a very high-tech device, it is rightly considered the “brain” of any computer. In one of the previous articles, we examined in detail the device of the central processing unit (CPU) of a computer. But, like any other component, the central processor has many parameters. And today, I propose to consider in detail the characteristics of the central processor.

Process technology

So, the technical process. Modern processors consist of a huge number of transistors placed on a small silicon crystal. The more transistors, the more powerful the resulting processor. It is possible to achieve a high mounting density due to the multilayer structure of the finished processor chip. The process is very similar to photolithography (when a photographic film is developed, light passes through the negative and creates an image on photo paper).

Modern technology allows you to create transistors measuring only 22 nanometers or even less! In comparison, the thickness of a human hair is about 50,000 nm. Over time, the manufacturing process will only decrease, which will create even more powerful CPUs, this trend is already evident. The smaller the process technology, the more transistors can be placed on one chip, and the more powerful the processor will end up, like this.

Architecture

The architecture directly determines the internal design of the processor (crystal circuit). Within the framework of one architecture, processors can have various characteristics: cache (more about this below), process technology, etc. Typically, such processors (with the same architecture, but with different characteristics) are said to have different cores. According to the established tradition, CPU manufacturers give different names to the cores to make it easier to navigate.

For example, CPU Intel Core microarchitectures were produced with different cores: Conroe, Merom, Kentsfield, Wolfdale, Yorkfield and others. The microprocessor core determines its 3 most important characteristics: clock frequency, FSB bus frequency and socket (connector). In addition, the kernels themselves can be refined many times, this is called “revisions” (stepping). In the process of such improvements, defects or weaknesses in the design are correct, and heat and energy consumption are reduced.

Kernels

The number of computing cores is another characteristic, the larger it is, the better, therefore. All existing processor manufacturing companies have long taken the path of increasing the number of cores placed on a single chip. Today it is already difficult to find models with less than two cores. Multicore – as a way to increase productivity is recognized as the most promising direction of the development of processors.

However, it is important to understand that the efficiency (performance) of the cores of various CPU models can vary significantly. In addition, not all existing applications (especially old ones) are optimized for working with many cores. And by default, they can use only one of them. And since many multi-core CPU clock speeds of each core are lower than that of single-core models, in such applications there may even be a decrease in performance.

Before continuing, I would like to talk a little about the main manufacturers of central microprocessors. Oddly enough, there are only 2 of them – Intel and AMD (just like the left and right stick “Twix”). And although these two giants, according to various estimates, own about 92% of all processors manufactured today. The market shares of these companies are not at all equal, as it might seem – Intel owns about 75-80%. The remaining 8% of products are highly specialized CPUs, as, for example, for mobile devices.

Multithreading

Since we are talking about kernels, it will not be out of place to mention such a thing as “multithreading”. The number of processor cores and the number of threads do not have to match. So, for example, the famous Intel Core i7 microprocessor with Hyper-Threading technology has 4 cores on board. But it works in 8 threads – which gives it a very good performance, even greater than that of some 6-core competitors.

Multithreading, in the case of modern 4-core CPU these are 8 threads, allows you to conditionally divide the application processing into 2 parts. That is, both parts of the application are executed by all the cores simultaneously (in parallel, if you want). This technology allows you to significantly increase performance in some specific applications that are “sharpened”. Or in other words, optimized for this technology.

In the case of older applications, or simply not optimized for multithreading, the opposite effect may be observed – a decrease in performance. Therefore, the BIOS of the motherboard provides the function of disabling hyperthreading at the processor when you need it. Multithreading will be very useful when rendering video or archiving a large amount of data.

CPU frequency

CPU clock frequency – the number of operations (clock cycles) per unit of time, and more specifically – per second. This parameter goes “hand in hand” with another equally important characteristic – the FSB bus frequency, which will be discussed below, and directly depends on it. The higher the frequency of the CPU – the more efficient it is. However, such a relationship can be traced only within the framework of one “line” (or in a different way – the lineup, as, for example, all CPU intel core 2 quad). Since in addition to the clock frequency, performance affects other parameters.

FSB bus frequency. This bus is a set of signal lines through which data enters and leaves the microprocessor. The frequency of this bus is proportional to the clock frequency of the processor, namely, the higher the frequency of the bus, the higher the processor frequency in the end. By the way, some novice (and not only) overclockers use this technique, but namely, they also raise the FSB bus frequency (“overclock” it), thereby increasing the processor clock speed.

There are several directions of “overclocking” a computer processor, it is possible to overclock “over a bus”, “over a factor”, “over a voltage”, etc. Overclocking “over a bus” is fraught with the fact that at the same time some other hardware “accelerates” along with the processor computer, including RAM, which may stop working when the threshold is exceeded the maximum working frequency of the memory. Also, if you do not specifically fix PCI slots in the BIOS, then the video card, sata (hard disks) and network card can “fail”.

Cache

Since the processor is very “closely” communicating with RAM, sometimes it can stand idle, waiting for data from it. A cache is a block of very fast RAM that is located right on the processor core. It acts as a buffer between RAM and the processor itself, instantly recording and transmitting information to it. There are several “levels” of such memory: the first level cache L1, L2 and L3. The cache of the first level is considered the fastest and outperforms conventional RAM in speed.

The consequence of using the cache is to increase performance. The larger the cache size at any level, the better. However, the L1 cache of the first level, as a rule, has a small amount (by modern standards) – only up to 128 kb. L2 cache of the second level performs all the same operations as L1, however, it has worse performance, but a large amount (up to 16 MB).

The larger the cache size, the more data can be written to it, however, the slower the processor will “get it” from there. Therefore, they came up with a separation of levels. I think you already guessed that the L3 cache will have the largest volume of all and the worst performance. But in fact, the L3 cache is not found in all processors, but only in the most powerful expensive solutions, as well as in server versions where it is really needed. Most processors have only two cache levels, which, however, are enough.

Heat dissipation

TDP (W) – an indicator characterizing the heat release (heating) of the processor during its operation. By TPD, you can indirectly judge the energy consumption of CPU. But do not equate them to each other, as it often happens. Because the power consumption of the processor is also measured in “W”. But the processor cannot emit as much energy in the form of heat as it was brought to it. And even more so – give up more energy, that is, generate it. Therefore, TDP will always be less by a few watts.

In the case of my processor (core quad q8400), the TDP is 95 watts and the power consumption is 136 watts. The TDP value is greatly affected by the process technology and the frequency of the processor core (to a lesser extent). The larger the process technology (nm), the stronger the processor will heat up. The same is true for frequency. TDP is also needed in order to assess how much power the cooler needs to be installed in the system in order to provide effective cooling.

Graphics card in the processor

In addition to the usual several cores, one can sometimes find another “core” responsible only for displaying the image on the monitor. That is, a miniature “video card” located directly inside the CPU. As a rule, they are equipped with all the “top” processors and most processors in the middle price segment.

Of course, the performance of such video cores can not be compared with full-fledged video cards. But it’s quite suitable for surfing the Internet and watching movies. They usually come with office computers of various organizations, laptops and netbooks, which saves on the purchase of a separate discrete (full) video card.

There is also such a connection. Usually, the more expensive the processor, the more productive the video core is installed in it. In the most powerful models (core i7, for example), the power of the graphic core is so high that it allows you to play modern games at medium, medium-low graphics settings, which in level corresponds to some budget video cards.

With all this, in the process of constructing a picture, a part of the processing power is taken from the processor and a certain amount of RAM is reserved as video memory.

Socket

Socket CPU is a connector (socket) on the motherboard of the computer into which the processor is installed. Accordingly, this “socket” should be designed to install a processor of a certain size (length, width) with a certain number of contacts on the bottom. If you plan to upgrade your computer (put a more powerful processor), be sure to see what kind of microprocessors your motherboard supports in general.

More about what a processor socket is already mentioned earlier, so we will not dwell on this here in detail. The same article briefly mentioned that a socket, among other things, affects the type of RAM (ddr2 or ddr3) that can be installed on the motherboard. For example, the previously ubiquitous LGA 775 socket only supported ddr2 type RAM. In general, different sockets correspond to different types of processors.

AMD makes sockets with a long “support”. In other words, each new generation of their processors does not always require a switch to another socket. With Intel, things are exactly the opposite. Almost every new generation of processors is released under a completely different socket. Which inevitably means changing the motherboard as well.

K

The presence of this letter in the name of the processor indicates the presence of overclocking potential. In other words, an unlocked multiplier is already coming from the factory in such a processor. This allows you to “overclock” the CPU without raising the FSB bus frequency. But only by choosing the multiplication factor (multiplier). In most CPU (not K) multipliers are locked at the kernel level. In K-series models, you are free to choose the multiplier value through the computer BIOS. Thereby overclocking the processor ONLY, and not the rest of the hardware.

Leave a Comment