Hi, Gypsyms! People's belief says that the grass of a neighbor is always greener, and computers who buy meticulous entrepreneurs for their needs are more reliable than and more productive than the model armed with marketing in retail. The whole cast of enthusiasts hunts the server components and the corporate class iron performance. We understand whether major organizations are really splashing in IT-Paradise, or did the guy have created an idol from nothing?

There are no obstacles, especially if these obstacles are erected by cunning marketers who divided everything electronic devices On corporate and CONSHUMERS! Because even in the media with an advertisement about the mysterious "user experience", the developers of software and iron are pronounced, they say, "the camera of this smartphone provides professional quality of images!", Yes, and another stamp about professionals who do not use nonsense, exploited for a long time. And if you look for the notorious "professional technique" and the quality of services, then it is better to ask the iron and methods of servicing the corporate class, right?

The motives that are guided by the restless enthusiasts lie on the surface - let the CONSHUMER equipment and develops beyree at the expense of buyers' appetites, "hardened by battle" components of the corporate class will clearly be more reliable, and in the secondary market - also cheaper. Play somehow geeks on video cards for workstations, they collect the mighty and "eternal" home PC with a server stuffing! Because it makes sense to try happiness?

And the Tolika of this very point in such a venture, of course, is, but with the acquisition of corporate "attributes" under household conditions it is possible to "enlighten" and, at best, overpay for unclaimed functionality, and in the worst - to go into minus in comparison with options, Available to retail buyer. We understand that it consists of tricks in the use of iron designed for corporations.

Server - also game. Intel Xeon in PC

The first thing that adherents of technologies love to use from the corporate segment are server processors. Not exotic, but the most "understandable", that is, on the basis of the X86 architecture. This pleasure is not cheap, therefore the "zoneonodes", conventionally speaking, include two camps with some different landmarks in the construction of PCs:

Xeon - initially not for games and "races" in benchmarks, but sometimes it is helpful

Enthusiasts aimed at High-END Accessories. This is a level when large-scale versions Intel Core The i7 is already lacking, and when looking at the LGA-2011 platform (any of the generations), the minds come to mind that the "super-shock" Core I7 offers "the same eggs", only in smaller quantities and without acceleration.

Because, since we are talking about the price, happened in the history of the moments when the eight-year Xeon turned out to be a third cheaper and significantly "colder" than the 6-nuclear Core i7 Extreme Edition. For example, it was so after the debut of Intel Haswell-E chips in 2014 - firstly, the difference in price between the six-core Core i7-5960x and the "civilian" four-sidelnist I7-4790K was a pitiful 15%. And secondly, the Junior Server Eight-core Xeon E5-2609 V4 cost about 30% cheaper than the candidate from the Haswell-E camp. At the same time, in contrast to the "simply" Core i7 in Xeon below the level of TDP and there is no useless for enthusiasts integrated into the processor of the graph.

At the same time, the L3 cache in all three models docked tons, and the frequency, even lower in Xeon, but the beliefs "are not unnecessary in the work of the nucleus" and "very soon games are optimized in such a way that they work quickly on 8 and more nuclei" do not give Economic fans of the speed of rest, after which the hot guys send the younger versions of Xeon to the Intel X99 chipset and ... do not recognize anyone how things are in games.

Therefore, the four nuclei diluted with Hyper-Threading almost always turn out to be more efficient in games than eight low-frequency "pots" in Xeon, which can not even dispel (a blocked multiplier, incomplete acceleration over the tire).

"Kulibins" who wanted to upgrade the old platform at minimum cost. For example, acquire an old core processor 2 Duo is not an old Quad, but a much coolest and high-frequency four-core Xeon X5460, which, with a simple adapter, can be installed not to the server motherboard with Socket 771, and in the "civil" for Socket 775.

The main thing in such a scenario is to be abandoned with high-quality cooling (servers "stones" firm TDP of about 120 W instead of 95 watts in standard four-core), but as a result, such an upgrade with a very old platform to "tolerant old" justifies itself, especially since on some Mattags The processor can be overclocked up to 4 GHz.

And after all, "Zyon" have the advantages with which they compensate for their multi-core inhretory in games! For example, the ability to make multiprocessor configurations with which video / music / music encoding / photo and CAD modeling occurs much faster than in the top Core i7 Extreme. Support for register memory with ECC, for example, allows you to correct the errors "on the fly", and this is notified with a large aptime (the server!). Support for the "horse" volumes of RAM and a huge number of nuclei will also come to the yard when the server needs to process incoming connections as quickly as possible. But all this is almost useless in the home PC.

And it is useful for him - a lot of high frequency nuclei. If these conditions are observed, the processor itself is compatible with platforms LGA 2011 or LGA 2011-3 and costs cheaper than "simply" Core i7 is the meaning in its acquisition. Otherwise, it is better to either bypass mass four-in-yidelines about eight threads, or design a workstation for specific use scenarios (rendering, coding).

High frequency Intel Xeon. (If they are cheaper by mainstream CPUs) can be a good help not only in the work, but also in games (Source: Ferra.ru)

Kosite frags on a workstation with novidia chaknut drivers

If using the server processor you can play rather contrary than thanks installed hardwareThe graphics that should be used for video modulation or design is historically steep in gaming disciplines. In the confrontation of AMD and NVIDIA, even the scenarios of "uncomplicated use" of video sources have always been different: "red" gamers' video cards have also recently been to grab from miners, and Nvidia Quadro, so historically, persuaded to retrain the game video card.

Professional video card NVIDIA Quadro much more productive than its game congor

And Quadro for these purposes is quite suitable - the fact is that GEFORCE gaming is most often a professional video card with partially disconnected conveyors graphic processor (from marketing considerations before chip rejection) at a more affordable price. For example, a new professional video card QUADRO P6000 contains the most "full" version of the GP102 graphics chip and for this reason, it takes a cool gamers' GeForce 1080 by almost 20%, and the mighty Titan X on the basis of the same Pascal architecture invariably leaves behind.

In general, among the fans of NVIDIA video cards, a corporate sport has long been formed - approach using the GEFORCE hardware modification to Quadro (for example, GTX 680 to the QUADRO K5000 analogue in terms of performance), and game lovers, on the contrary, crossed the hedgehog with the "pick" drivers and Forcing professional video cards to work faster in post-stores / cruising / stalks. "Play as conceived" such activity does not allow, but the persistence of enthusiasts can only be envied.

In mobile workstations, almost every NVIDIA Quadro video card has an amusing pattern: any NVIDIA Quadro mobile video accelerator is equal to the game GeForce class below in gamers and a pair of levels of coolest gaming geforce in CAD disciplines.

Mobile NVIDIA Quadro performance compared to GeForce analogues (Source: MSI.com)

For example, Quadro M2000M in games shows itself at the level GeForce GTX. 960m, but as soon as it comes to modeling, "bounces" in the results to GeForce GTX 980M. Approximately the same ratio is true and in the case of other quad models: M5000M competes with GTX 980M in games, and M1000M competes from 950m in games.

NVIDIA Quadro M6000 in comparison with the fastest game video cards
(Source: techgage.com

Children ice cream, lady - Flowers: Priorities in corporate memory and drives

Server rAM Not compatible with motherboards in the home PC is not because someone decided "called" to finite buyers. Just the server RAM is located a little differently - it contains a register between chips and a memory controller in order to reduce electrical load on the controller and be able to set more modules in one memory channel.

In other words, additional chips and ability to automatically recognize and correct errors improves the fault tolerance of this type of memory, but also increases its value. In short, do not be surprised if you find that even low-frequency (by the standards of DDR4 standard), the modules will be 50% and more expensive than their "household" analogues - inhuman requirements in endurance in round-the-clock systems included systems have noticeably changed the server RAM. In everyday use, it will not be faster, neither more efficiently "civilian" counterparts, therefore, for high performance it is worth contacting gamerskits - for example, Hyperx Savage, if you need a convenient memory for gamers, and Hyperx Predator, if you want to squeeze out of the subsystem RAM maximum. For standard frequencies, the budget Kingston Valueram is well suited - reliable, once installed and forgotten.

The server processor in the home PC can be useful, but instead of register memory it is better to purchase a standard DDR3 / DDR4 kit.

The SSD of the corporate class also underwent tuning towards reliability - in them, for example, it is possible to flexibly manage the backup volume for the needs of the controller. The greater the volume - the lower the wear of the cells and the higher the durability of the drive. And a huge amount of algorithms, effective in heavy working conditions, especially on part of the data saving in case the drive turns off in emergency mode. Refassed to a minimal delay in multi-user access mode Firmware and a struggle for stable performance even with the freelancer of record and read operations. Such a load of a home computer is not worried, even if "torturing" SSD torrents. On the other hand, the record holders in typical operations industrial SSD are also not - typical SATA drives are faster than "morally", from the point of view of memory volume, than completely exhausted the number of overwriting cycles available for cells - a long-term comparative test with the participation of Hyperx models. And the speed records at the same level of reliability have long moved to the NVME interface databases, which are implemented in one of the new-fashioned form factors "top" PCI-Express. The Kingston / Hyperx model line "King Mountains" was and remains Predator SSD PCI-E.

Winning in durability when buying an SSD corporate class will not compare with the joy of speed of Gamer's PCI-E drive

If it is impossible, but I really want - you can

The corporate class iron is not so different from the "civilian" analogues to recognize it unsuitable for work as a home PC, simply always need to proceed from whether it is worth the heater. Because the situation is as follows:

Buy a platform in which register memory with error correction (ECC) is used for home - a bad idea. Excessive durability does not compensate for expensive components and medium (in comparison with gameimascular analogs) the level of performance will not be delight, especially since the price of server memory is noticeably higher than on the average DDR3 / DDR4 module.

Corporate Class Drives in home computer Need if you are paranoid, extremely worried about the safety of data in case of electricity interruptions and experiences regarding the reliability of modern SSD in general. Organizations oriented drives will allow you to "unscrew the maximum" validity rates so that the soul is calm.

Server processor for games ... Curious and quite effective idea, but only in the case when it comes to more cheap (in comparison with mainstream analogues) and, most importantly, the high-frequency model. Or about the old computer upgrade to the server CPU " small blood", That is, almost for nothing. And yes, ideally, the platform should be borrowed from the "ordinary" extreme-series of mass processors.

Professional video cards perfectly copble not only with modeling, but also with games. But it should be remembered that in mobile workstations (with a "strangled" TDP), a professional middle-class video surveyor will be able to compete in gameiman disciplines only with the gaming video cards of the budget class. And the desktop professional video cards, in turn, although fast in all scenarios of work, are harderfully costly, and they certainly do not fit the role of economy option for "work and play".

Be that as it may, it is impossible to save on high-quality and fast RAM ... But today - you can! We remind you that from February 2 to 20 to all memory sets

Hello, friends! In this article we tried to answer numerous questions related to RAM. ? How to find out what kind of RAM has I installed and how much? How to choose the right memory for your computer. How to find out if your RAM is working in two-channel mode or not? What is better to buy, one memory bar with a volume of 8GB DDR3 or two strips of 4 GB each? And finally.

• If you are interested, or, too, read our articles.

1. Hello Admin, one of my friend asks to establish him more rapid memory. The properties of the computer show the volume of 2 GB. Turned off the computer, opened system unit, there is one straightfight of RAM, took it out, and there is no designation on it. What is interesting, failed to determine the model of the motherboard. The computer was bought for a long time, accordingly, the question arose - and how to find out the type of RAM that he needed? After all, the RAM differs in type, frequency and timing.
2. Hello everyone! I wanted to buy the RAM, removed the cover of the system unit, took out the bar of the RAM and can not decipher the information applied on it, is simply written there serial number And that's it. It is completely incomprehensible at what frequency it works and what kind of it is type, DDR3 or DDR2. How to distinguish DDR3 memory from DDR2, how do they differ externally?
3. In my system unit, one DDR3-1600 DDR3-1600 DRAGE plate with a volume of 4 GB, I want to put another plank, too, with a volume of 4 GB, but operating at a higher frequency DDR3-1866. My computer will work normally, and most importantly in two-channel mode?
   My buddy installed three different volume and frequency of the RAM. Is it allowed? But what is strange, his computer works fine!
4. Tell me how to check, is my random memory in two-channel mode or not? And what conditions are needed so that my memory worked in two-channel mode. Same volume? Same frequency or identical timings? How faster the computer works in two-channel mode, rather than in one-channel. It is said that there is still a three-channel mode.
5. What will be better to work, two strips of 4 GB RAM in two-channel mode or one plank, but the volume of 8 GB, respectively, the memory mode will be single-channel?

To find out all the information about the RAM module, it needs to be carefully considered, usually the manufacturer lakes the Radio frequency, volume and type of RAM. If there is no such information on the module, then you need to know everything about the motherboard and installed processorSometimes this action turns into an integer investigation.

1. Important comments: Friends, do not forget that all new processors Intel Core i3, Intel Core i5, Intel Core i7 RAM controller is located in the processor itself (before this, the northern bridge of motherboard) and memory modules are now directly controlled by the processor itself, the most applies to the latest AMD processors.
2. This means that it does not matter what frequency of RAM supports your motherboard. It is important which frequency of RAM supports your processor. If your computer has a processorIntel Core i3., Intel Core i5., Intel Core i7., Formally supported processor data memory standards: PC3-8500 (DDR3-1066 MHz), PC3-10600 (DDR3-1333 MHz), PC3-12800 (DDR3-1600 MHz), It is at such frequencies Your RAM will work, even If the motherboard passport states that the motherboard can work with high frequency RAM PC3-19200 RAM (DDR3-2400 MHz).
3. Another thing if your processor with unlocked multiplier, that is, with the letter "k" at the end, for example, CPU Intel Core i7-4770K, 3.5 GHz.The unlocked multiplier refers to the fact that a computer with such a processor can be installed as the highest frequency memory planks, for example DDR3- 1866 MHz or DDR3-2400 MHz, such a processor can be accelerated and the RAM will operate in its frequency 2400 MHz . If you install the RAM barDDR3-1866 MHz or DDR3-2400 MHz into a computer with a regular processor, that is, with blocked multiplier without a letter" K "at the end, for exampleIntel Core i7-3770, 3.9GHz, then such a bar will work at best at the frequencyDDR3-1600 MHz, and at worst - the computer will not be loaded. Therefore, buy RAM of the appropriate processor.
4. Concerning processorsAMD. recent years, then they work with memoryPC3-10600 (DDR3-1333 MHz).

How to find out all the information about RAM you have installed?
First, on the player of the RAM must be all the information you are interested in, only you need to read correctly. I do not argue, there are a plank of memory, on which there is practically nothing, but we will cope with them too.
For example, take the HYNIX RAM bar, it has such information: 4 GB PC3 - 12800.

What denotes the following:

first, the volume of 4 GB,

second, 1RX8 - RANX - a memory area created by several or all memory module chips, 1RX8 are one-sided wounds, and 2RX8-Double-sided memory.

As we see, it does not say on this bar that it is DDR2 or DDR3, but indicated bandwidth PC3-12800. PC3 - The designation of peak bandwidth belonging only to the type DDR3 (RAM of DDR2, the designation will be PC2, for example PC2-6400).

This means that our Plank RAM produced by Hynix has a type of DDR3 and has a PC3-12800 bandwidth. If the bandwidth of 12800 is divided into eight and it turns out 1600. That is, this memory plan of the DDR3 type, operates at a frequency of 1600 MHz.

Read everything related to DDR2 and DDR3 RAM on the site

http://ru.wikipedia.org/wiki/ddr3 And you will become understandable.

Take another RAM module - Crucial 4GB DDR3 1333 (PC3 - 10600). This denotes the following: volume 4 GB, the type of memory DDR3, the frequency of 1333 MHz, the bandwidth of PC3-10600 is still indicated.

Take another bar - Patriot 1GB PC2 - 6400.

Manufacturer Patriot, 1 GB volume, PC2 bandwidth - 6400. PC2 - Tick bandwidth designation belonging only to DDR2 type (DDR3 RAM 31 The designation will be PC3, for example PC3-12800). The bandwidth of 6400 is divided by eight and it turns out 800. That is, this DDR2 memory bar is operating at 800 MHz.

Another plank - Kingston KHX6400D2 LL / 1G
Manufacturer Kingston, bandwidth 6400, type DDR2, volume of 1 GB. We divide the bandwidth to 8, we get the frequency of 800 MHz.
But there is a RAM on this bar yet important information , She has a non-standard nutrition voltage: 2.0 V - exhibited in the BIOS manually.

RAM modules differ from each other by the size of the contact sites and the location of the cuts. Using the cutout, you will not be able to set the RAM module into an unbelievable slot for it. For example, the DDR3 memory bar cannot be installed in the DDR2 slot.

Everything is clearly visible according to this scheme.

Sometimes there will be no understandable information on the RAM module, except for the name of the module itself. And the module cannot be removed, as it is warranty. But in the title, you can understand what kind of memory it is. for example

Kingston KHX1600.C9D3 X2K2 / 8G X, all this means:

KHX 1600 -\u003e Run running at a frequency of 1600 MHz

C9 -\u003e Timings (delays) 9-9-9

D3 -\u003e DDR3 RAM Type

8G x -\u003e 4 GB volume.

You can simply dial the name of the module in search engines and you will learn all the information about it.
For example, the information of the AIDA64 program about my RAM. Kingston Hyperx RAM modules are set to RAM 2 and 4 slots, DDR3 memory type, 1600 MHz frequency
Dimm2: Kingston Hyperx KHX1600C9D3 / 4GX DDR3-1600 DDR3 SDRAM
Dimm4: Kingston Hyperx KHX1600C9D3 / 4GX DDR3-1600 DDR3 SDRAM

Is it possible to install the RAM bar in different frequency?

RAM frequency does not have to match. The motherboard will display the frequency for all installed rackels of RAM on the slower module. But I want to say that often the computer with the straps of different frequency works unstable.

We draw a simple experiment. For example, we take my computer, it has two identical Kingston Hyperx RAM module, the DDR3 memory type, the frequency of 1600 MHz.

If you run the AIDA64 program in my Windows 8, it will show such information (see the next syringe). That is the program Aida64 shows simple specifications each of the RAM planks, in our case, both planks have a frequency1600 MHz. But the programAIDA64 does not show at what frequency now the RAM strips are operating, it is necessary to look in another program calledCPU-Z.

If run free program CPU-Z and go to the MEMORY tab, then it will show which frequency your RAM scrolls work. My memory works in two-channel DUAL mode, the frequency is 800 MHz, since the DDR3 memory, then its effective (double) speed of 1600 MHz. So my RAM strips work on the frequency for which they are 1600 MHz. But what will happen if next to their RAM Runs operating at frequency1600 MHz i will install another bar with frequency1333 MHz!?

We set into my system unit an additional DDR3 memory bar running at a lower frequency of 1333 MHz.

We look at what the AIDA64 shows, the program shows that an additional plank is installed with a volume of 4 GB, the frequency is 1333 MHz.

Now launch the CPU-Z program and let's see at what frequency all three planks work. As we see the frequency of 668.7 MHz, since the memory DDR3, then its effective (double) speed of 1333 MHz.

That is, the motherboard automatically put the frequency of all RAM on the slower module 1333 MHz.

Is it possible to install RAM strips with frequency more than supported motherboard The most important thing is that the frequency of RAM is supported by your motherboard and processor (there are information about the processors at the beginning of the article). For example, take maternal asus board P8Z77-V LX, it supports modules operating at 1600/1333 MHz in nominal mode and 2400/2200/213/2000/1866/1800 MHz in acceleration. All this can be found in the passport on the motherboard or on the official website http://www.asus.com

Installing the RAM bar in the computer with a frequency greater than supports the motherboard is not desirable. For example, if your motherboard supports the maximum frequency of 1600 MHz, and you installed the RAM module operating at a frequency of 1866 on the computer, then at best this module will work at a smaller frequency of 1600 MHz, and in the worst case the module will work on its frequency 1866 MHz, but the computer will periodically reboot oryou will receive when downloading a computer blue screenIn this case, you will have to enter the BIOS and manually set the frequency of RAM in 1600 MHz.

Timing (signal delays) Determine how often the processor can refer to RAM if you have a four-core processor and it has a large second-level cache, then too large timings are not terrible, since the processor is less likely to turn to RAM. Is it possible to install the RAM bar in the computer with different timings? Tyigns also do not have to coincide. The motherboard automatically will display timings for all slats along the slower module.

What conditions are needed so that my memory worked in two-channel mode Before buying RAM, it is necessary to study the maximum information about the motherboard. All information about your motherboard can be found from the leadership of the attached to it when buying. If the leadership is lost, you need to go to the official website of your motherboard. You will also use the article "How to find out the model and all information about your motherboard"
Most often, in our time there are motherboards supporting the following operational memory modes. Dual Mode (two-channel mode, meets most often) - With careful consideration of the motherboard, you can see that RAM slots are painted in different colors. This is done specifically and means that the motherboard supports two-channel operational memory mode. That is, two modules of RAM are specifically selected with the same characteristics (frequency, timings) and the same volume and are set to the same RAM Slots in the same color.

If there is one RAM plan on your computer, but the motherboard supports two-channel mode, you can purchase exactly the same in frequency and volume of the RAM bar and install both planks into the same DIMM slots.

Is there an advantage of a two-digital mode before one-channel

For normal work You will not noticize the difference on the computer, but when working in applications, actively using RAM, such as Adobe Premiere pro. (Video installation), (Canopus) Procoder (video encoding), Photoshop (work with images), games, difference can be felt.

Note: Some motherboards will operate in two-channel mode, even if you set the DIMM slots in the same color different memory modules. For example, in the first DIMM slot you install the 512MB module, and the third slot of the bar volume of 1GB. The motherboard activates the two-channel mode for the entire volume of the first plank 512MB, and for the second plank (which is interesting), it is also 512MB, and the remaining 512MB the second plank will work in one-channel mode.

How to find out if my RAM is working in two-channel mode or not? Download the free CPU-Z program and go to the MEMORY tab, see the Channel parameter in our case - Dual, then the RAM is working in two-channel mode. If the Channels - Single parameter, then the RAM is working in one-channel mode.

Triple Mode (three-channel mode, rarely meets) - You can set from three to six memory modules. What will be better to work, two 4 GB RAM strips in two-channel mode or one plank, but the volume of 8 GB in one-channel mode?

My opinion, with usual work on the computer, will be the same to work, I personally did not notice a special difference. I worked for a long time on a computer with one big strip RAM and the performance was the same as on exactly the same computer with two straightfight RAM working in two-channel mode. A survey of friends and familiar sysadminov strengthened me in this opinion. But when working with programs actively using RAM, such as Adobe Premiere Pro, Canopus ProCoder, Photoshop, games, a computer with two RAM strips will work faster.

Is it possible to install several different in the frequency and volume of RAM on the frequency and volume?

Of course you can, but not desirable. The computer will work more stable if it will be implemented that the operational memory mode will be implemented, which is recommended in the motherboard passport. For example, two-channel mode.

27. 06.2018

Dmitry Vasiairova Blog.

What is the difference between the server RAM from the usual - is there a difference?

Good day, my dear readers and I am glad to communicate with you again. Today, it is impossible to be called popular, because it seems to be ordinary computers. But in fact, the question is what the server's operational memory differs from the usual RAM has become increasingly exciting ordinary users.

I would associate it with unsuccessful attempts Upgrades based on the logical assumption that iron for equipment operating around the clock will be high quality and reliable.

But on the fact, the server hardware is components with a narrow specialization. Therefore, let's understand.

Between the server and the usual working or gaming computer There is a significant difference due to responsibility for the tasks solved. Therefore, the requirements for the installed gland are in the root differ.

For server equipment, working 24 hours daily, it should not be just reliable, but fault tolerant. In the server DDR memory this is provided in different ways.

Hardware support

In particular, the servers establishes, differing from the usual presence of an additional chip that performs the role of buffer. It is smaller in size, placed in the center of the plank, so you can easily distinguish such a module. Usually, 1 buffer is installed for every 8 ordinary chips. What is it for?

The fact is that on modern motherboardsah the RAM controller is an integral component of the processor. But since, while simultaneously handling several memory modules, it is subjected to serious current loads (due to the change in the electrical capacity of the chip in the process of "write-reading"), then he needs reliable protection. This feature and performs the buffer of the module of the server register memory. Do not be it, the server processor at intensive work could easily fail.

Software

In the process of reading information with memory chips, an error due to the effects of external factors may occur. Do not be surprised, neutrons of cosmic and powerful electromagnetic radiation are able to easily change the status of the memory bit.

To minimize the consequences of such a situation, the ECC error code adjustment function is used, which also uses some of the individual modifications of ordinary memory. The algorithm used in it is capable of mathematical methods for processing a digital code to detect and correct errors. Is it worth saying how important it is for stable server work?

Immediately I want to draw the attention of readers to marking server memory. You may also know that the Modules with ECC are denoted by the "E" letter. But it does not mean that such a module is a server.

Remember: Only register memory can be server, and the ECC is already its mandatory component. It is indicated by the server memory with letters in the labeling "R" or "REG", which means "REGISTERED". The type of such RAM itself is called FB-DIMM (FULL BUFFERED).

It is also worth adding that the failover of the server RAM is provided not only by the above methods. In addition, it passes special testing, imitating long-term operation (heating up to 100 ° C) under intense load. After that, memory modules are checked for compatibility with different software and hardware server platforms. This allows you to identify defective modules in a short time. If their number is greater than the laid (2 strips of 10 000 pcs.), Then the whole game is braked.

Distinguishing

As you can see, the reliability of server RAM simply amazes and it is quite natural that some users want to use it on the usual computer. But, my dear friends, there are several nuances here and I want you to know about them:

• The exchange of information through the buffer will require additional work cycles from the processor, in addition, the ESS algorithm is used as it also requires additional time for processing. As a result, the server memory demonstrates a much smaller speed;

• You understand perfectly well that the availability of additional chips and high quality requirements / reliability of the product significantly affect the final cost of the product. Therefore, the price of server memory is much higher than usual;
• And finally, the main information for those who wish to find out: will the register memory work on the usual motherboard? Will be. But not every. And server and gaming support various planks. They can have the ability to work with RAM buffer. This technology can significantly increase the amount of RAM without creating additional load on the processor. Therefore, always specify the technical characteristics of your motherboard and, perhaps you will manage to install reliable server memory on PC.

What is the difference between the server RAM from the usual you now know. The differences are not so much, but they are very significant. On this end my story and say goodbye to you. I hope soon to please you with new interesting articles.

See you and all you benefits!

More and more people face the problem of incompatibility of RAM with a computer. Install memory, and it does not work and the computer does not turn on. Many users simply do not know that there are several types of memory and which type approaches their computer, and which is not. In this manual, I will briefly deal from personal experience About RAM and where each applies.

You do not know what it means U. in labeling of RAM, which means E., What means R. or F.? These letters denote the type of memory - U. (Unbuffered, non-buffered), E. (Memory with error correction, ECC), R. (register memory, REGISTERED), F. (FB-DIMM, Fully Buffered Dimm - fully buffered DIMM). Now consider all these types more.

Memory types used in computers:

1. Nebuperized memory . Normal memory For ordinary desktop computers, it is also called udimm. On the memory bar, as a rule, there are 2, 4, 8 or 16 memory chips from one or two sides. This memory, the marking usually ends with the letter U (unbuffered) or at all without a letter, for example DDR2 PC-6400, DDR2 PC-6400U, DDR3 PC-8500U or DDR3 PC-10600. And the memory for laptops marking is ends with the letter S, apparently this is a reduction from SO-DIMM, for example DDR2 PC-6400S. Photo of non-buffered memory can be seen below.

2. Memory with error correction (Memory with ECC.). Ordinary non-buffered error correction memory. Such memory is usually placed in branded (branded) computers sold in Europe (not server), the plus of this memory is its great reliability when working. Most errors when working in memory can be fixed during operation, even if they appear without losing data. Usually, one or 2 microcircuits is added on each bar, one or 2 microcircuits is added. Most of the usual computers (non-servers) and motherboards can work with ECC memory. In such a memory, the marking usually ends with the letter E (ECC), for example DDR2 PC-4200E, DDR2 PC-6400E, DDR3 PC-8500E or DDR3 PC-10600E. Photo of nebufurized memory C ECC can be seen below.

Difference memory with ECC and memory without ECC can be seen in the photo:

At least most of the boards sold and support this memory, but compatibility with a specific board and processor is better to know in advance before buying. From the personal experience of 90-95% of motherboards and processors can work with ECC memory. From those that cannot work: cards on Intel G31 chipsets, Intel G33, Intel G41, Intel G43, Intel 865pe. All motherboards and processors starting from the first generation Intel Core can all work with ECC memory and from motherboards does not depend. Under AMD processors In general, almost all motherboards can work with ECC memory, with the exception of cases of individual incompatibility (this happens in the rarest cases).

3. Register memory (REGISTERED). Server memory type. Usually O. always issued with ECC (error correction) and with chip "buffer". Microcircuit "Buffer" allows you to enlarge maximum amount Memory planks that can be connected to the bus without overloading it, but this is already extra data, we will not delve into the theory. Recently, the concepts of buffered and the register almost do not distinguish. If you extend: register memory \u003d buffered. This memory works only on server motherboards Capable to work with the memory of the black chip "buffer" chip.

Usually, 9, 18 or 36 memory chips and 1, 2, 2 or 4 buffer chips are installed on the register memory planets with ECC (they usually in the center, differ in memory microcircuit). This memory marked as a rule ends with the letter R (registered), for example, DDR2 PC-4200R, DDR2 PC-6400R, DDR3 PC-8500R or DDR3 PC-10600R. Still in the labeling of the register (server) (buffered) memory is usually there is a reduction in the word registered - REG. Photo buffered (register) memory C ECC can be seen below.

Remember! Register memory with ECC with 100% probability does not work on ordinary motherboards. It works only on servers!

4. FB-Dimm. Fully buffered dimm (Completely buffered DIMM) is a computer memory standard that is used to increase reliability, speed, and memory subsystem density. In traditional standards, the data line memory standards are connected from the memory controller directly to the data lines of each DRAM module (sometimes through buffer registers, by one register chip per 1-2 memory chip). With an increase in the width of the channel or the data rate, the quality of the signal on the bus deteriorates, the tire layout is complicated. This limits the speed and memory density. FB-DIMM uses another approach to solve these problems. This further development of the idea of \u200b\u200bregistered modules - Advanced Memory Buffer exercises unnecessary address signals, but also data, and uses the serial bus to the memory controller instead of parallel.

The FB-DIMM module has 240 contacts and the same length with other DDR DIMM modules, but differs in the form of protrusions. Suitable only for server platforms.

FB-DIMM specifications, like other memory standards, published by JEDEC.

Intel. used FB-DIMM memory in systems with Xeon series 5000 and 5100 series and newer (2006-2008). FB-DIMM memory is supported by server chipsets 5000, 5100, 5400, 7300; Only with Xeon processors based on Core microarchitecture (Socket LGA771).

In September 2006, AMD also refused plans to use FB-DIMM memory.

If you find it difficult to choose a memory for your computer, then specify the seller who informs it the motherboard model and the processor model.

P.S.: Recently, another cheap and interesting type Memory - I call it " Chinese fake"Who else has not come across - I'll tell you. This is such a memory that can always be recognized by its contacts, usually they are oxidized and even if they clear them, then for the month or two they again oxidize, become muddy, dirty and memory can get closer or Do not work at all. Gold on contacts of this memory will not even smell. Another distinction of this memory from the original is that it works on certain motherboards or processors, for example, only on AMD, or just strictly on some chipsets. And the list These chipsets are very small. What is the secret of this "memory" yet it is not yet clear to me, but many buy - it is 40-50% less than a similar one. And that the most amazing, new "Chinese fake" is usually cheaper than the original memory: ) I will not tell the reliability and durability of work, here everything is clear.

By virtue of its specifics, the server theme is a relatively rare guest on the covers of periodic IT-editions. If you ask the quantitative ratio of the existing "server" and "desktop" information on the Internet, then you can get approximately 1:20. As for even more subtle matters, such as the organization of data storage system on the enterprise's scale servers, it is very difficult to get such knowledge. Today we will touch on, perhaps, one of the most "quiet", but still those present areas of the market of professional components, namely server memory.

Since the very phrase "server memory" is rather an abstract jargonal expression containing too many different values, we decided to combine the "general education" part with the definition of the economic feasibility of the existence of this direction as such.

What is server memory? To lead our conversation more items, to start it would be nice to deal with the terminology and explain what is meant by this phrase in this publication. So, server memory (hereinafter JV) is the memory control modules and error correction, as well as additional functionality to ensure the increasing stability (register buffered memory), created according to the standards that are used in desktop products and certified for use in servers from A-brands. We will not claim that this definition is listed in the textbook, but the essence seems to us, it reflects.

The second question that I would like to discuss is - how the server memory differs from the desktop, in addition to the above additional bits, registers and buffers?

Differences between server and desktop memory

Manufacturers

The number of brands in this market is much more modest than on the "desktop" field. In order not to get involved in who, what and for whom it makes, we break the company, producing the joint venture, into several subcategories.

A-brands - Memory created under specific servers manufacturers with applying their unique marking on chips and modules. All the developers of the first echelon (HP, Dell, etc.) typically use a similar approach, allowing to increase total profits. Square special attention to the one who specifically releases one or another module and / or a chip, in this case, perhaps, it is not worth it (they can be any company who won a tender or has direct contracts). The most important thing is that you need to know: the modules sold, for example, under the Dell brand name, with Dell servers are accurately compatible, and with a 100% probability will work in them.

Chip A-Brands (Memory Chip manufacturers and modules on them) - Companies engaged in the issue of both microcircuits and memory modules under their own brand. This category includes Micron, Samsung, Hynix, Quimonda (former Infineon). By and large, it is they who regulate the memory market as a whole, as the total produces about 70% DRAM chip. All of the above firms have product line of server modules of any standards. Of course, a rigid binding to production allows you to have more competitive prices compared to companies that are focused exclusively on server brands, but in this case there is another side of the medal - complexity with certification. For example, changing the series or generations of microcircuits leads to a change in marking (sometimes only chips, and sometimes modules), which requires new tests from servers manufacturers to issue a conclusion about whether it is suitable if new Memory For operation in their systems or not. There are situations where two different product in their characteristics are in essence under identical marking (and both original), which gives a lot of headache to servers.

Modular A-brands (manufacturers of modules on third-party chips) - The most common category. Among them can be called such well-known names as Kingston, Corsair, Transcend, Apacer. Actually, such companies in relation to the joint venture are often called "test", because their engineers are most of the time they are engaged in testing modules for performance with existing on sale server platforms. As a result, a situation arises, in many respects similar to the joint venture from server A-brands, in addition, such firms have much less problems associated with marking. Therefore, the end user or server collector can easily get information that the module with "such" marking is suitable for use in the "such" server, and it does not matter, on the basis of the microcircuit of which manufacturer it is made.

All three approaches have their own positive and negative sidesBut in the near future it is not necessary to wait for any changes in the presence or alignment of brands in the GP market.

fault tolerance

In relation to the joint venture, rather desktop, term, as "reliability", is usually replaced by "fault tolerance"that more accurately reflects the meaning. Since such equipment should be non-stop working 99.9% of the time since its commissioning, during production and testing uses significantly more stringent approaches than in the manufacture of desktop products.

For example, the technology of "artificial aging" provides the detection of production flaws during a two-day period - during testing, server modules are heated to 100 ° C, which makes it possible to quickly lead them to a state corresponding to two months of operation. Next, in the set of tests that heat the memory subsystem, it is checked for compatibility of modules with various server platforms, which is still somewhere day. As a result, the joint venture with a percentage of marriage is about 0.02% (one module for five thousand).

In the same section, it is worth saying a few words about such a concept as a "successful model", again come from the desktop world. The famous fact that there are "successful" video cards, equally well-working on any platforms, "successful" hard drives, compatible with most controllers, and "successful" memory modules that ensure stable operation with almost all motherboards. As for the joint venture, everything is exactly the opposite. The main criterion, which is guided by the manufacturer, is the lack of "unsuccessful" models, because the quality-created SP Planck should work everywhere and always without any reservations. So, if the product from the brand x did not work with the brand Y server, then such a joint venture is most likely not allowed to the conveyor until they determine the cause. Of course, to afford a similar approach for components for mass computers can no one.

Criterias of choice

The final buyer, choosing memory when assembling or upgrading its computer, as a rule, is guided by the following criteria: a brand (here is a guarantee), the price, test results. In other words, it is important for him who the manufacturer, which you can get timings and frequencies, and how much for all this will have to pay. If you rise to the step above and see what the principles are guided by the collector, then the price immediately comes to first place. It, in view of a very tough competition in the market, usually chooses the cheapest option from trouble-free.

The following is the following integrators, as a rule, are primarily guided by the experience of operation of certain modules in certain platforms and focus on one (maximum two) suppliers capable of strictly performing conditions such as supply stability and a rapid solution of technical issues with the manufacturer In case of their occurrence. The price although is definitely an important component, but it is already far from the defining factor. If the joint venture from one brand will be 20% more expensive than the other, but it will also provide better compatibility, then the choice will fall, most likely on it.

What determines the demand for memory: the opinions of server manufacturers

Evgeny Bobruichko

eVEREST Server Equipment Manager

In accordance with the Company's policies, we try to offer an optimal solution for specific applications, be sure to with a small reserve for the future. If the client "overgrown" is its configuration, then in many cases it is more expedient to acquire a more modern and powerful server, and the old translate to other, less responsible and labor-intensive tasks.

Services for the modernization of servers use in our company very low demand (about 2% of systems are subjected to it). After all, the emerging problem of improving the memory subsystem indicates either a significant increase in the client streaming tasks (for example, due to an increase in state), or an integrator error that suggested an inefficient solution.

As for the types of memory, the FB-DIMM has excellent performance in streaming data, which is very in demand when working with the database (especially OLAP), and also allows you to install a significant amount of RAM to the server. On the other hand, it has high latency, which is not always good. For DDR2, this indicator is smaller, but it has a lower bandwidth. It is to combat high latency in Intel who have a good "weapon" - a big and common for two cache kernels from WoodCrest.

In general, it seems to me that the FB-DIMM has a very good prospects, since DDR2, due to the parallel data transfer method, as the frequency increases, approaches the ceiling. However, I do not exclude that at first Intel can make a budget version of system logic on DDR2.

Andrei Tishchenko

managing Entry

In my opinion, the modernization of the servers is ungrateful, because the entire component base is observed and the simple addition of many problems cannot be solved. Most of our customers are growing, constantly updating their server park, old models they adapt to less resource-intensive applications without upgrading. Therefore, requests for the installation of additional memory for us are rare, such sales do not exceed 2-3% of the total supply.

I must say that there is no competition between DDR2 and FB-DIMM as such - there is a planned transition of two main chipamers to a new architecture. From who this period will make it easier to, the popularity of one or another memory depends. Intel has made a big step forward by swaying the limitations of the shared processor tire and reducing the power consumption of new CPUs. Nevertheless, the rate on FB-DIMM in the near future can play against the company. In comparison with DDR2, its indicators are worse: higher latency, power consumption and heat dissipation, and the cost of modules is about 10%.

The AMD with the transition to Socket 1207 (Rev F) and the adaptation of the memory controller on the kernel under the DDR2 retains the ability to work with memory at the core frequency, using the most mass standard, scalability of multiprocessor multi-core platforms. In the future, the company involves the transition to the memory of FB-DIMM, but he sees his trump cards in another. Figuratively speaking, if Intel constantly stimulates developers of new types of memory and supports high DDR-DDR2-FB-DIMM evolution rates, then AMD is greaterly focused on compliance with the price / performance / power consumption.

Igor Przhegarlinsky

commercial Director of Onyx

If a 32-bit operating system was installed on most of the servers, which imposed restrictions on the amount of memory to 4 GB, then with the transition to 64-bit OS, the limit will increase to 32 GB (for 1-4 processor servers on windows database Server 2003 Standard Edition).

With the release of new intel processors (Dempsey and Woodcrest) and AMD (Socket F) to the market are the types of memory FB-DIMM and DDR2-667 ECC REG. Both chipmeters have long refused to support a single standard, and competition between memory types as a result will be concerned about the competition between the microcircuit producers.

The main volume of server memory today is used in new systems - the proportion of modules sold for modernization does not exceed 5%.

Performance

Highly for a long time SPE significantly lagged in their high-speed characteristics from desktop. It is enough to remember that the DDR400 standard that has managed to become massive for desktop systems and laptops, the server systems came only with the appearance of Opteron. Even in modern chipsets for powerful workstations, the register buffered DDR266 continued to use for a long time.

The next jump in professional systems on intel database It was the use of DDR2-400 - this is despite the fact that in desktops, the start of the DDR2 began with a mark of 533 MHz. For reference, we note: For servers based on Itanium 2, only DDR200 was used with 128-bit access to provide the required bandwidth. The reasons for which platform creators chose more low frequencies, understandable: improving the reliability and reduction of the memory controller load integrated with the main set of logic.

Today, the server equipment require maximum speedThat forces the use of the most modern standards, not inferior, and sometimes superior to existing for desktop systems. Please note the specification of the latest server platforms - the total performance of the Northern Bridge tires in Intel chipsets and integrated controllers from Opteron can translate for a fantastic mark of 30 Gbps. And most importantly, at such speeds it is necessary to provide a 24/7 operation mode in the performance of very resource-intensive applications.

Perhaps the situation is even more difficult with the memory subsystem. Today's standards are FB-DIMM 667 MHz for Intel and register buffered DDR2-667 with dual parity check for Opteron. The amount of memory to ensure comfortable work is constantly increasing, and, as is well known, the likelihood of an error in RAM is growing exponentially with increasing volume. As a result, the module manufacturers and, first of all, the CP microcircuits have to search for ways to ensure reliability not less than at the previously used DDR200 / 266 standards, as well as in the increased volumes that are currently up to 32 GB, huge frequencies in 667 MHz and preservation The cost difference is not more than 40% compared with desktop modules.

Price

Although the SP market is not subject to strong fluctuations, which are often observed in the desktop segment, but at the same time it is sufficiently dynamic, and such methods of conducting a competitive struggle are also not excluded here, as dumping, marketing actions, OEM supplies at discounted prices and T . D. Today, the cost of the joint venture modules differs on average by 20-50% of those similar in volume of desktops. On the one hand, it is not enough, on the other, it is possible to recall a more complex technical implementation (additional chips to monitor parity and buffering), the need for tests of tests after the product output from the conveyor, certification from the servers manufacturer and, of course, a lifelong warranty. As a result, it turns out that the Earnings of the Vendor on the modules of the joint venture is hardly much more than on desktop "Planks". And the modern market requires a constant price reduction: it suffices to remember that FB-Dimm began to be cheapest literally from the day of their appearance in free sale. So, if at first the cost of FB-DIMM is roughly doubled the joint venture of DDR2 and DDR standards, now this difference is much more modest and is about 30-40%.

What determines the demand for memory: the opinions of suppliers

Elena Krivoshenko

head of the Sales Department of Memory Modules and Flash Products of Kiev-Tech

According to our data, the volume of the Ukrainian segment of server memory is 10-13% of the memory market as a whole, which in financial terms is approximately 5-7 million dollars per year.

Over the past months of 2006, the strategy of sales of server modules has changed somewhat. DDR and DDR2 are sold at approximately equal volumes, and recently a lot of DDR modules have enough to upgrade servers. As for FB-DIMM, today their share in the overall sales structure is no more than 2%.

In quantitative terms, the server memory market, according to our forecasts, will increase. It can be assumed that in the next six months, the distribution structure will shift towards DDR2. Regarding FB-DIMM modules, I believe that their share will become gradually increasing, although in the next six months it is hardly essential. Solutions using this type Memory is still very expensive - primarily due to the cost of processors.

Dmitry Borovsky

tNG General Manager

Due to the complex political situation, this year, the corporate IT market segment was quite strongly slowed down. This led to a decrease in the supply of comprehensive solutions, the basis of which servers are. There is hope that at the end of the year the situation will change, but it is difficult to ensure that the demand for components, including server memory, will be fully satisfied in the "hot" season. Based on this, the annual volume of the Ukrainian server's memory market can be estimated only approximately - according to our forecasts, it will be from 40 to 60 thousand modules, and in financial terms - from 4 to 5 million dollars.

I will note that in 2006, sales of DDR333 server memory decreased dramatically, almost the same happens with DDR400 - their share is now less than 5%. The main product today is the DDR2-400 modules, and the demand for FB-DIMM is due to the presence in the market of motherboards for them. The supply of FB-DIMM began only in the third quarter, and while in our sales they take less than 5% (however, the trend towards the growth of this indicator is obvious).

There is no particular difference between trends in the domestic and international markets. For systems in which one of the most important factors is fault tolerance, manufacturers prefer the memory of well-proven vendors. Perhaps the only difference is that major world brands are mainly used in their products server memory from the "first hands" - from Samsung, Hynix, Micron. The local market marks a significant demand for modules from Corsair, Kingstone, etc.

Andrey Semenovsky

managing company NeBesa.

Server memory segment is just part of the servers market, however, due to the high cost of modules, playing a more prominent role than memory in the desktop market. If the desktop is important for the release of successful memory models, around which marketing is built, for servers - lack of unsuccessful. Problems with the availability and quality of server memory can lead to costs multiplied by the price of the modules, and therefore responsibility for the delivery time, the stated compatibility of the models and the quality in this case is an order of magnitude higher. And considering also low profitability of server memory sales, you can imagine how heavy this segment in terms of business.

Of the features of the domestic market, I will note a rather high potential of Ukrainian server collectors. The amount of memory consumption is about 6-8 thousand modules for the quarter, and the average price of the plank is close to $ 150.

Changes in the sales structure depend on those in model row Platform manufacturers (primarily Intel and AMD). By the time the FB-DIMM exit, the sales ratio of DDR2 and DDR was 70:30. With the advent of completely buffered memory, the transition began: the demand for platforms with support for DDR2 is noticeably declining, and from FB-DIMM - not too active due to enough high Price and non-ability. It is noteworthy that during this period, sales of DDR - their share in the structure of the supply approached 50%.

Viktor Shcherbian

head of the Sales Department of ASBIS Ukraine

Exit new server operating systems Allowed to significantly increase the limit amounts of memory in the system (Early versions of the OS had significant limitations on this parameter). Therefore, sales of modules both in a barn and monetary expression have grown. According to our data, the server memory segment is about 5% of the entire market for this product.

The demand for DDR with the appearance of new server platforms has decreased markedly. The largest sales are now on the DDR2 modules, the implementation of FB-DIMM is growing, although the share of the latter is still insignificant. According to my estimates, the sales volumes of both types of memory by the end of the year are likely to be equal.

In general, the servers every year require all the large scalability, flexibility, capacity, speed, as well as the reliability and safety of investments. Based on new technologies, server memory manufacturers produce modules that allow you to achieve these improvements.

Ukrainian realities

According to available information from integrators for the year in Ukraine, about 10,000 servers are sold in Ukraine (clarify: not computers that perform the role of the server, namely systems using the joint venture). Let even this figure is too optimistic, but the point is that the amount of SP modules of the SP is approximately 20,000 "planks", among which about a third goes to the modernization of previously collected systems. For averaging, suppose that the cost of one module is $ 100, therefore, the total volume of the joint venture in our country is estimated by about $ 2 million. And the number of firms offering such products does not exceed five, and they work mainly with such vendors like Kingston, SAMSUNG and HYNIX, of which the first two are about 70% of the Ukrainian market.

Probably the information provided in this article for someone will seem too "basic", but we believe that for the first topic, the number dedicated to the server memory, it clearly does not hurt. And more in detail with the technical aspects of this type of products, we offer you to get acquainted in the following material.