Internet Windows Android

Amd dual core c 50 description. What does the best eSports store in Ukraine offer? Working with the Over Drive program

If the overclocking, and even the processor, then it will start again: CPU-Z, Prime-95 and Linpak ... And these are programs that do not actually take part in the "overclocking". But, in fact, it turned out to be a little easier with AMD. Much easier.

The Canadian company AMD, that is, the company itself, produces one such program. It is absolutely free. From there, you can overclock AMD processor (starting from AM-2 socket), on any "motherboard", regardless of manufacturer ... Change all values, test the correctness of overclocking, watch real frequencies, test performance. That is, one program (with one window from several tabs) - will replace the typical "set" of utilities. But for everyone, no one forbids testing "stability" with Prime, as well as evaluating performance after overclocking with Linpak. We repeat once again - the program runs freely on all motherboards (with socket from AM2 and higher, and AMD chipset from 7xx). It is also called, simply: AMD OverDrive.

Warning

Any change in the values ​​of clock frequencies that goes beyond the limits established in the documentation (as well as overestimation of the supply voltages) violates the license agreement and voids the final warranty. After "overclocking", any device automatically loses its warranty. You will carry out all actions at your own risk.

Now - about less sad

The program allows you to "change" almost everything that can be changed: the frequency of HyperTarnsport, PCI-e and PCI buses, even (attention!) - memory timings. Well, and voltages (and all this - with temperature monitoring in a constant mode). A multi-core amd processor can be overclocked separately for each core ... In a word, having AMD OverDrive installed, there is no need to go into the BIOS.

Official Requirements

Chipsets supported: AMD Hudson-D3, 990X, 990FX, 970, 890GX, 890FX, 890G, 790FX, 790GX, 790X, 785G, 780G, 770.

Your motherboard chipset is not listed? Most likely, it really is not supported (including, this applies to 760G, 740G, 780V).

The program is downloaded here:

http://download.amd.com/Desktop/aod_setup_4.2.3.exe. At the time of writing this review, version was 4.2.3 (which is discussed below).

Preparatory actions

Where should a person go before going to first grade? That's right, preparatory. So it is here:

  1. Cool-n-Quiet driver, if it was installed - leave it: this is AMD Processor Driver for Windows, let it remain.
  2. Go to BIOS and turn off forcibly:
  • Cool 'n' Quiet (in Disable);
  • C1E (in Disable);
  • Spread Spectrum (in Disable);
  • Smart CPU Fan Control (in Disable).

When exiting the BIOS, be sure to save the changes. Boot the OS.

Note: another name for C1E is Enhanced Halt State. It is impossible to provide a detailed manual here, since all motherboards are different (if we don’t know what is where, we read the instruction booklet for setting up this BIOS).

Actually, the system is now ready to install and run "Over Drive". But first, a couple more words.

Can the processor be overclocked in this system?

Take a look at the power consumption graph. It concerns just overclocking (that is, consumption - before and after this action):

This is the power consumed by the processor alone (in watts). Immediately, a couple of questions appear: will your power supply "pull"? And what about the processor cooler? At AMD, as a rule, all boxed coolers are designed to work in "standard" modes (that is, the cooler is almost at its limit even without overclocking). If you can answer yes to both questions, go to the next step.

Note: 248 watts here falls on a 12-volt line (that is, the current through it is 20.7 amperes, while not many power supplies can "boast" a value higher than 20).

Working with the Over Drive program

For a start - a short educational program.

  • Processor frequency is the frequency of the CPU core at which the processor is executing instructions.
  • HyperTransport frequency: the frequency of the interface between the processor and the northbridge. Usually - equal to the frequency of the north bridge (but - should not exceed it).
  • Northbridge (NB) frequency: For processors, increasing the northbridge frequency results in faster memory controller (and L3 cache) speeds. This frequency should not be lower than the HyperTransport frequency, although you can make it much higher.
  • Memory frequency: the operating frequency (in megahertz) at which the memory functions. It should be remembered that the physical frequency is 2 times less than the "effective" one.
  • Finally, the base frequency: as you can see, all frequencies are calculated from the base frequency (by multiplying or dividing it).
  • CPU clock speed = CPU multiplier * base;
  • Northbridge frequency (aka L3 frequency in AMD) = Northbridge multiplier * base;
  • HyperTransport frequency = HyperTransport multiplier * base;
  • Memory frequency = memory multiplier * base.

We launch the Over Drive program. In the first window - click "OK" right away:

Thus, the user agreed to the responsibility (associated with the undesirable consequences of "overclocking"). The main program window will appear after this:

As you can see, all frequencies currently set in the computer are shown (HyperTransport frequency is in the right column, and HT ref. Is sort of "base").

Why are there so many "multipliers"? Isn't it easier to overclock the computer at once with the base frequency?

The fact is that two more are connected with the "base" - these are the frequencies of the computer buses, PCI and PCI-Express. With an increase in the PCI frequency, many devices built into the board can work unstable (and this is already observed when less than 10% is added to the "standard" values).

This program for overclocking the amd processor allows you to monitor temperatures (everything). Go to the "Status Monitor" tab (the second in a row):

Here we see the temperatures of only the processor cores (in the last line). Choosing "Board Status" and "GPU status", we get a similar "screen" for the motherboard and video. The fact is that the latest version supports overclocking of the video accelerator built into the processor (and in the previous versions - only in the chipset, and also Side Port). That is, the video temperature must also be controlled ... But we overclock the processor.

Go to the "Performance Control" tab (third from the top).

This is the main overclocking window. But now the tab is in the "for beginners" mode. We go to the last ("Preference"):

Here (the “Settings” tab) - instead of “Novice Mode” we select, as in the figure (“Advanced Mode”). If you return to the previous tab, it will look like this:

Well, finally! You can freely change all frequencies (that is, all multipliers), including even the "base" frequency (denoted as "HT ref."):

Note: as you can see, the northbridge multiplier (NB) is absent. The NB frequency, in fact, increases "automatically" with the HyperTransport frequency change (it cannot be less, can it?).

As you can see, the overclocking margin for HyperTransport (hence, NB, and most importantly - L3 cache) is quite small. The base frequency, however, cannot be "pulled up" to very high values ​​(even at 220 MHz, something can "hang", including: sound, network ...). So, first of all, they usually play around with the processor's Core Multiplier.

You can activate the changes with the "Apply" button:

After that, it is better to check if the overclocking has led to instability (“Stability Test” tab). Well, the real performance can be assessed in the "Benchmark").

CPU overclocking technology

  1. Increase the processor multiplier (let it be +1 or 2). It was 15 - now it is 17. Click on "Apply".
  2. Turn on "Stability Test". If it passes - we run to the "Status Monitor" tab (write down the temperature).

If everything suits you (if the processor warms up no higher than up to 70-75 degrees), you can increase the frequency even more. That is, steps 1. and 2. are repeated, but only until the appearance of “undesirable” temperature values ​​(or, “failure” of the “Stability Test”).

Thus, we overclocked the processor with only one multiplier.

Here, also - "Stability Test" after each change. Limit - when one of the devices (integrated into the motherboard) starts to malfunction. The point is to reach the maximum possible CPU frequency with a lower multiplier (a gradual increase in the "base").

In general, overclocking at the "base frequency" requires some skill.

And last but not least (the third stage, so to speak) - you can also "increase" the "HT Multiplier" multiplier. That will lead to overclocking the L3 cache (and even more CPU heating). After you finish overclocking, run the Stability Test. Always (when changing something, including one other than the CPU multiplier) - see the temperatures (not only of the processor, but also of the motherboard) given on the “Status Monitor” tab.

After "overclocking", the program itself can be closed. All settings will remain (to "lower" them - run the program again). You do not need to restart your computer (and, even after a restart, the changes will remain in effect).

Additionally

We only "overclocked" the processor. The weak link in the system will remain memory. It is also possible to overclock it, the "Memory" tab is used for this:

But this is more difficult than overclocking the CPU, since "stable" RAM overclocking is associated with the selection of timings (delays during switching). Of course, you can immediately increase them by a couple of values, but then - all the same, it is better to choose carefully.

The name is lit "red" - the value takes effect only after a reboot. "Memory Clock" translates into English as "Memory Clock".

Note: for memory class DDR-3 (and 2), the physical frequency (displayed by the program) - refers to "effective" as one to two.

It may be strange, but the memory voltage is regulated in the same place as all the others (in the “Clock / Voltage” tab). Their values ​​are increased, if in another way it does not work. Anyway, overclocking by changing voltages is recommended "last."

After overclocking the system, do not be lazy to run the "Stability Test". At very high multiplier values ​​(more than + 20% to the "standard" values), it is better to watch the temperature immediately after pressing the "Apply" button (continuously, 8-10 minutes). If there is overheating, immediately change the value to the "previous" one.

We need competent, that is, "stable" overclocking, and we do not want "overheating shutdown". Is not it?

Well, how much can you "overclock" a certain processor? First, all non-Black Edition processors will not allow you to change the multiplier (Core Multiplier). This means that you can overclock the Core (core) only a little bit, that is, with the "base" frequency. And more - nothing, in theory. But, it is this "overclocking" that increases the performance of the system "as a whole", in a proportional number of times.

If the user nevertheless decides to configure the memory through the program, he must first enter the BIOS. To set memory timings (only, manually):

By default, they are always "Auto", so this step (at the preparatory stage) is required.

Explanation: the computer takes memory timings from the SPD of the memory itself (at each new boot of the PC, if the value in the BIOS is "auto"). In turn, the SPD contains the values ​​"recommended" by the manufacturer. Instead of "auto" mode, you need to set each timing value in "explicit" form (and how to do it - well, at least the same as in the SPD).

That is, we take, go in, change (instead of “Auto”, it becomes “5”, then “5”, and so on, according to the displayed data from the SPD). SPD translates as: "sequential prediction detector", in general, the name does not reflect the meaning (in Russian, it would be more like "memory ROM").

There are a lot of values, but changing them is real (in the BIOS shown here - only 9, then 5 more). Everything should work out ...

Overclocking statistics

Let's take and consider now the randomly selected results from "Overclockers.ru" (from the statistics on overclocking of the "lightest" family in this sense - Propus, aka Athlon-II X4).

First result: 3667 MHz (282 "base" * 13.0). Cooler - BOX. Raising the voltage was still used (the actual Vcore was about 1.5 volts). Conclusion: as you can see, the base frequency is quite overclocked. Cooler - no need to change. We used a very "not weak" motherboard (ASUS M4A78LT-M), with a "not weak" power system. Nominal CPU frequency: 200 * 13.0.

Result of the second: 3510 MHz (234 * 15.0). Vcore = 1.416 (that is, not too high). And this is a stable overclocking (it seems that the "base" one could not be increased any further), but the board was also "not simple" - ASrock 870 Extreme3 ​​(cooler - BOX). Normal mode: 200 * 15.5.

Third result: 3510 MHz (260 * 13.5). Sometimes the "basic" one can still be overclocked (on the ASUS M4A77T board). The voltage is almost "standard" (1.5 Volts), but the cooler was needed absolutely "not BOX" (Cooler Master Hyper 212 Plus). Normal mode: 200 * 15.0. The temperature of all Cores "at maximum", and - in full processor load mode, did not exceed 50!

In the first example, the temperature is 62 Gy. C, in the second - 50.

Advanced Clock Calibration (ACC)

How to overclock an AMD processor - we've covered in some detail. But, there is one more function that you need to know about. An "ultra-precise" frequency matching function that is performed automatically (called ACC).

ACC is only present on boards with 750 or higher southbridge. ACC itself can be enabled both in the program and inside the BIOS (in both cases, a reboot is needed).

Why are we talking about this here? For a 45nm Phenom II processor, the best option is to disable ACC (after all, AMD claims this feature is in the processor die). The same is true for any CPU from those. process "no older". For older processors (Phenom and Athlon 65nm), ACC should be set to Auto. + 2% to + 4% frequency gain is guaranteed.

So go to our “favorite” tab (Performance Control) and check the value.

What can influence the "success" of overclocking?

At the very beginning, it was already said that during overclocking, the processor requires more power. At AMD, most desktop processors fit into the 95W package. But this does not mean that the power (both consumed and allocated) must be at this limit.

By the way, the situation has not improved lately. AMD FX processors, despite using the 32nm process technology, remained at about the same level (the TDP value did not drop below 95).

For overclocking, "three" devices are important: the CPU power supply system (on the motherboard), the power supply unit (as mentioned above), and the processor cooler.

This "set" must be "balanced", that is, all components must fully comply with the requirements of the others. The user probably guesses that there is no point in installing a "cool" motherboard if the power supply unit "does not pull" even half of the total power. In general, 20 Amperes is the "minimum" of the power supply, for its 12 Volt line (240 watts, but there are also big requirements). Gluttony, that is, the power of the processor, goes nonlinearly with increasing frequencies. At the beginning of the review, we showed (how much the 965 "eats"). The load also increases with increasing Vcore supply voltages.

All this power has to be “dissipated” (all this is released in the form of heat on the CPU itself). For Athlon II - more often a BOX cooler is enough, but about more "powerful" processors - don't say so ... Here we are talking about overclocking, of course.

All these requirements are very important. However, overclocking is a lottery, the final result will depend on the processor instance. All the same "strapping" - will only help to reveal the potential. Do not believe too much the statistics (as well as reviews), where 45-nm "stones" - exceed the 4.0 Gigahertz limit. Instances are different (Core is chasing - but "cache" is not chasing), the options are different, and what to overclock (and whether it is necessary) is up to the user himself.

Overclocking results

We will not write about performance, about its growth along with "overclocking". The real speed of work - indeed, changes, and changes for the better (but non-linearly with the frequency itself).

Let's look at a couple of cases here. That is, the consequences (while not too desirable).

The user has not "overclocked" the new processor. At the end of the warranty period, this was "fixed", and almost immediately. Everything was done correctly (the maximum frequency was found, etc.).

The PC itself worked in this mode for 2 months. Well, then - he stopped (as it were, broke). What is not a reason for panic?

The problem was the same - only in the connector on the board (it was highly oxidized, as a result of which, 12V was not supplied to the processor). The rest is in order, it turned out after replacing the connector. However, in the "normal" mode, the computer would continue to work, nothing would have to be changed (just the connector, as luck would have it, was 4-pin).

A common defect can also be considered a tap of the board transistor in the CPU power circuit (power transistors on the "motherboard"). If before overclocking everything seems to work, then the user himself conscientiously “turns on” all tests that cause maximum “power” (and the computer takes it and “turns off” during these tests) ... after such a defect, the motherboard will not recover. Keeping track of the temperature value - it turns out that it is impossible (well, there are no such sensors on the "motherboard"). S&M is considered a powerful test for "overheating", while Prime95 finds errors faster than others.

That is, in "overclocking" - errors are possible. Outgoing from the "overclocking". The likelihood of which is the lower, the higher the quality of the rest of the "hardware" (as it was considered: motherboard, power supply unit, and so on). And the quality is also more expensive. Maybe for the same amount - take a faster processor ...

Whether it makes sense to overclock is up to the user himself. What to overclock and what to check - you make the choice yourself.

The information given here should be enough for "basic" overclocking. More fine tuning of "hardware" - requires qualifications.

AMD manufactures highly upgradeable processors. In fact, the CPUs from this manufacturer only operate at 50-70% of their actual capacity. This is done so that the processor lasts as long as possible and does not overheat during operation on devices with a poor cooling system.

There are two main ways to increase the clock speed of the CPU and speed up the processing of data by the computer:

  • With the help of special software. Recommended for less experienced users. AMD itself is engaged in development and support. In this case, you can see all the changes immediately in the software interface and in the system performance. The main disadvantage of this method: there is a certain probability that the changes will not be applied.
  • With the help of BIOS. Better for more advanced users as all changes that are made in this environment greatly affect the operation of the PC. The standard BIOS interface on many motherboards is entirely or mostly in English, and all control is done using the keyboard. Also, the very convenience of using such an interface leaves much to be desired.

Regardless of which method is chosen, it is necessary to find out whether the processor is suitable for this procedure and, if so, what is its limit.

Find out the characteristics

There are many programs to view the characteristics of the CPU and its cores. In this case, let's look at how to find out the "suitability" for overclocking using AIDA64:

  1. Run the program, click on the icon "Computer"... It can be found either on the left side of the window, or in the center. Then go to "Sensors"... Their arrangement is the same as with "Computer".
  2. The window that opens contains all the data regarding the temperature of each core. For laptops, a temperature of 60 degrees or less is considered a normal indicator, for stationary computers 65-70.
  3. To find out the recommended frequency for overclocking, return to item "Computer" and go to "Overclocking"... There you can see the maximum percentage by which you can increase the frequency.
  4. Method 1: AMD OverDrive

    Method 2: SetFSB

    SetFSB is a versatile program that is equally suitable for overclocking processors from AMD and Intel. It is distributed free of charge in some regions (for residents of the Russian Federation, after the demo period you will have to pay $ 6) and has simple controls. However, there is no Russian language in the interface. Download and install this program and start overclocking:


    Method 3: Overclocking via BIOS

    If, for some reason, through the official program, as well as through a third-party program, it is not possible to improve the characteristics of the processor, then you can use the classic method - overclocking using the built-in BIOS functions.

    This method is suitable only for more or less experienced PC users, because the interface and controls in the BIOS can be too confusing, and some mistakes made in the process can disrupt the computer. If you are confident in yourself, then do the following manipulations:


    Overclocking any AMD processor is quite possible through a special program and does not require any deep knowledge. If all precautions are taken, and the processor is accelerated within reasonable limits, then nothing will threaten your computer.

The Zona51 online store is a unique place on the Ukrainian Internet, where professional gamers, enthusiasts, game lovers and other public, not indifferent to branded devices, as well as high-quality paraphernalia for PCs, laptops, tablets and smartphones, regularly visit.

A good gaming store is still a rarity in the electronic space of our country. Therefore, cybersportsmen have already appreciated the convenient portal with an impressive catalog of gaming devices and accessories for professionals in FPS, RPG, MOBA, PBEM and other genres.

Features of the assortment of goods for video games and e-sports in "Zone51"

Armchairs

The best e-sports store in Ukraine offers real gaming chairs from the world's leading manufacturers. A gamer sitting on an ordinary stool or chair is like a Formula 1 racer who risked rolling out a broken-down Zaporozhets on the track. First, it’s unsafe and terribly inconvenient. Secondly, the chances of winning are zero. No esports player has yet won a competition on a stool. Take pity on your health - make yourself a gift, after the appearance of which you will say thank you to yourself every day.

Keyboard

Want to level up your game? Go to an eSports online store for a gaming keyboard. Only in it you will find the necessary additional keys and functions that are so lacking in cheap peripherals not suited for gaming. Feel comfortable during protracted gaming battles. Customize the controls to your liking and forget about minor mechanical damage. Additional iron fasteners and elements increase the wear resistance of the professional device.

Headphones

One of the main tasks of an esports player is to hear everything that happens in the gaming space. For this purpose, the store for gamers sells headphones for professional players of different genres. For a high price, you get impeccable sound quality, a device that will serve you for many years and will not have a damaging effect on your hearing, even if you turn on the music at full power.

Mice

The mouse is the main weapon of the gamer. Therefore, every self-respecting e-sports online store sells these babes who bite a little with prices and impress with their technical capabilities. Not a single ordinary mouse has yet saved any virtual life. And if you want to outrun your opponent by a split second, do unique tricks and combinations, you cannot do without a professional manipulator.

Rugs

A good gaming surface is just as important as a professional mouse. In our eSports store, you will find specialized rugs designed for a specific type of game and manipulator. Easy glide or rough surface for precise cursor positioning - the choice is yours. All the rest was taken care of by the world's leading manufacturers.

We have representative offices in Kiev, Kharkov, Odessa and Lvov. But you will only need them if you have time to break away from your favorite game and personally honor us with your presence. Well, if you pass by, of course, it's a sin not to go to see the magnificent technological masterpieces that thousands of the best gamers of the solar system are armed with every day.

For all other cases, we have prompt delivery to any locality in Ukraine. The hardest part is making a choice. But we are confident that with the help of free consultations from our experienced sellers, you will find a solution to this difficult task.

Many computer users have heard that you can significantly improve the performance of your computer by overclocking its processor. In this article we will talk about how how to overclock an AMD processor, we will introduce you to the peculiarities of this operation.

As a rule, a newly purchased computer becomes obsolete in a year and a half due to the rapid development of modern technologies. Very soon after the purchase, it begins to fail to cope with new games that require large computing resources, to slow down. Overclocking the processor will extend the life of the computer, saving a significant amount on buying a new one, or on replacing its main parts (upgrading). In addition, some people use overclocking immediately after purchase, trying to increase its performance to the maximum, because in especially successful cases it can be increased by 30%.

Why is overclocking possible?

The fact is that AMD processors have a large technological reserve, built into them by the manufacturer for reliability. To understand how to overclock the amd processor, you will have to say a few words about its device. The processor operates at a certain frequency, which is set for it by the manufacturer. This frequency is obtained by multiplying the base frequency by an internal multiplier that the processor has and can be controlled from the BIOS. For some of them, this multiplier is locked, and such are not very suitable for overclocking, while for others, you can change it yourself. The base frequency is generated by a generator installed on the motherboard. The frequencies of this generator are also used to generate other frequencies necessary for the normal operation of the computer. This:

  • The frequency of the channel that connects the CPU and the northbridge. Typically this is 1GHz, 1.8GHz, or 2GHz. But in general, it should not be more than the frequency of the North Bridge. This channel is called HyperTransport.
  • The frequency of the North Bridge also depends on this generator, the frequencies of the memory controller and some others depend on the same frequency.
  • The frequency at which the RAM operates is also determined by this generator.

From this we can draw a simple conclusion - maximum computer overclocking is possible only when choosing components that function reliably in extreme conditions. First of all, these include the motherboard and RAM.

The question arises how to overclock amd phenom or athlon processor? There are two ways to do this - you can increase its multiplier, or you can increase the frequency of the base generator. Let's say our generator has a standard frequency of 200 MHz, and the processor multiplier is 14. Multiplying one by the other, we get 2800 MHz - the frequency at which the processor operates. By setting a multiplier of 17, we get a frequency of 3400 MHz. True, whether our processor will work at this frequency is a big question! The second way is to increase the frequency of the base oscillator. By increasing its frequency by 50 MHz, we will have a processor frequency of 3500 MHz (with a multiplier of 14), however, this will also increase the frequencies of all elements of the board, which depend on the generator.

Heat dissipation system

With an increase in frequency, the heat release of any element always increases and a limit comes when it refuses to work at a given frequency. In order to restore his working capacity, the voltage on him is increased. This, in turn, increases the heat it generates. Ohm's law says that increasing the voltage by 2 times increases the heat dissipation by 4 times. Hence the simple conclusion - in order to successfully overclock the amd processor with a hair dryer (athlon), you need to take care of its good cooling. Moreover, if overclocking is carried out through a generator, then the motherboard must also be cooled. For cooling, both coolers of increased performance and water cooling are used, and in extreme cases - liquid nitrogen.

Overclocking the processor

It can be done using the AMD OverDrive utility, which allows you to both overclock the processor and test its operation. This utility is produced by AMD, and is designed to facilitate this process.

But many users prefer to carry out such overclocking through the motherboard BIOS. True, this path requires some theoretical training and knowledge. You will also need a utility that will allow you to evaluate the result - this is CPU-Z, it will show the new processor frequency and Prime95 - a utility that allows you to evaluate the stability of the system under overclocking conditions, as well as some others - to control temperature and performance.

BIOS settings

Depending on the type of motherboard, the settings in the BIOS may change, but we recommend setting some of them like this:

  1. Select Disable for Cool 'n' Quiet.
  2. For C1E select Disable
  3. For Spread Spectrum select Disable
  4. Select Disable for Smart CPU Fan Control

You should also set your power plan to High Performance mode.

Remember that all actions to overclock the processor you perform solely at your own peril and risk!

Overclocking technique

It is recommended to overclock the amd athlon (phenom) processor by increasing its multiplier step by step by one notch. After each increase in the multiplier, you need to check the stability of the processor at the new frequency using the Prime95 utility, and if the test fails, make another attempt by increasing the voltage on the CPU by one step. After the test has been passed without errors at least three times in a row, you can increase the multiplier by one more level and try to pass the tests again. Acting in this way, you will find the value of the multiplier and voltage at which the processor will be stable, and the next increase in the multiplier should lead to the fact that the test will not pass. After this value of the multiplier and voltage is found, it is recommended, for permanent operation, to reduce them by one step. When overclocking, carefully control the temperature of the processor, it should not go beyond the limits set by the manufacturer.

If, by changing the value of the multiplier, it is not possible to obtain a high overclocking, then it is worth trying the second way - to increase it by increasing the frequency of the base generator.

In this short article, we have covered the very principle of how to overclock amd athlon and phenom processors, without dwelling on the details. For those who want to know more about this, there is a lot of literature, both in paper and electronic form.

What's interesting about the new platform?

The idea of ​​a single chip combining the functions of both a central processor and a graphics accelerator has been hovering over the computer systems market for a long time. However, for desktop or mobile computers, single-chip solutions did not exist until recently. Moreover, the PC architecture has traditionally provided for a large number of different chips: processor, video, chipset (two independent chips), very often - various peripheral controllers.

Meanwhile, the integration of as many system components as possible into a single chip promises significant advantages. If all the necessary computing units and controllers are in one chip, it is both cheaper and more efficient. The speed of interaction between components increases. The design of the boards is simplified, there is no need to interconnect several different chips with high-speed buses. In most cases, energy consumption and cost are reduced, and the cooling system becomes simpler and more efficient.

AMD has been working on component integration for quite some time now. One of the striking examples is the transfer of the RAM controller from the north bridge to the processor. However, the company took its most decisive step in 2006, in its largest merger deal with Canadian graphics chip and system logic maker ATI. The strategic goal of the merger was the development of a single integrated platform that would combine the functions of both a central processor and a graphics core (moreover, a full-fledged one, so that, for example, an additional chip would not be required to organize image display). The company named it APU (Accelerated Processing Unit). So great was the stake in creating an integrated platform that AMD even changed its own logo, adding the slogan "The Future is Fusion" to it. The first marketable product was supposed to appear already in 2010.

Another thing is that AMD very rarely manages to meet the deadlines or the declared functionality. I ran into this when AMD released the Puma platform. On paper, the characteristics looked very impressive, but in practice nothing interesting came of it. Not to mention the very interesting concept of XGP, which, it seems to me, AMD ruined with its own hands, failing to properly build the interaction between product manufacturers and consumers.

Unfortunately, the development of the APU only confirmed the general trend. After the acquisition of ATI, AMD added a new slogan to its name and, spreading its sails, moved on to create a single processor, and that was back in 2006. However, the development took so long that Intel's main competitor, having managed to swim in many other seas during this time, came to the finish line first. How did it happen? Moreover, the Arrandale platform (the first generation of Core i3-i5) with a very strange internal organization, where inside a single processor case there were two completely different cores, a CPU and a graphics controller, even made according to different technologies (32 and 45 nm, respectively), is already completing its life cycle, and a new generation of Sandy Bridge is entering the market with might and main, in which the blocks of the central processor and graphics controller are already organically integrated and united by a single bus.

And only here the AMD Brazos platform appears on the market with two processor options (codenamed Zacate and Ontario during development).

Does this mean AMD is late? It may not receive the status of a technology leader, but let's not exaggerate. After all, the new AMD platform that has entered the market is aimed at the segment of the market of low-performance devices: tablets, netbooks, ultraportable laptops. Intel is entering the market with powerful multi-core battleships with tremendous salvo performance. AMD offers a less productive, but at the same time very cost-effective solution for mobile and ultra-mobile solutions - which, I must say, are experiencing a real boom now. If the company manages to capture this growth and gain a foothold in the market (which, however, there are some doubts), it will be an undoubted success.

Indeed, Intel in this segment can only respond with the Atom platform, which is distinguished by both low performance and very weak functionality (moreover, in many cases the functionality is narrowed, as they say, "for political reasons"). For example, she still doesn't have an external digital video output and is unlikely to get one in the near future. Therefore, in order to get HDMI output and more or less decent graphics performance, you have to fuss with NVIDIA ION2, which in the current situation cannot be called anything other than a perversion (an external chip is "hung" on the PCIe 1x bus, in addition to the usual platform). You can read more about this in our material on the history of netbooks.

However, it should be noted that at least the netbook segment is very price-sensitive. Therefore, you can sell a lot of devices, but will you be able to get a big profit from this?

Technological aspects of APU

However, let's leave the conceptual reasoning for the end of the article and move on to analyzing the new AMD platform. Which, by the way, has already been repeatedly considered in our materials.

There are two APUs in the Brazos line, codenamed Ontario (9W) and Zacate (18W). Between themselves, they differ in clock frequency, 1 and 1.6 GHz, respectively. You can read more in our presentation of the architecture of the new AMD processors. It also describes the Bobcat core, on the basis of which the processors participating in today's testing are built.

After entering the market, the codenames are discarded, Ontario is now - Series C, Zacate - Series E. In total, four processors should enter the market, two in each line. They differ from each other in the number of cores - one or two. C-30 and C-50 are called for a 9-watt system and E-240 and E-350 for an 18-watt system, respectively. In mid-December, a preliminary performance review of AMD Zacate processors by Alexey Berillo was released, which describes the platform and conducts some preliminary tests.

In addition to the APU chip itself, the platform includes another hub, which is similar in functions to the traditional south bridge. In the current platform, this is a powerful and functional Hudson M1 chip, which, however, may turn out to be a little more power hungry than one would like for an ultra mobile platform. You can read more about its functionality in the corresponding review.

Finally, a material has recently been released, which compares the performance of the E-350 processor and its main competitor, Intel Atom, in real applications. The comparison is made using the example of desktop systems. On the one hand, it makes it possible to compare the performance of different solutions more clearly, on the other hand, a lot of interesting things remained outside the scope of the material, for example, energy consumption issues.

Well, we are moving on to the study of mobile processors. Today we have a summary material in which we will evaluate the performance of two chips at once - C-50 and E-350. And for comparison, we will take a wide variety of systems based on Intel processors from different lines.

Participant configuration

To begin with, let's define the test participants and their technical characteristics. In general, with the selection of configurations, there was some overlap, since, as it turned out, we had not yet tested a single Intel Atom-based netbook using the new method, and the netbook that we had impeded the process in every possible way (we never ran the test suite. succeeded). Besides, as it turned out, the test suite of applications runs on a netbook for about a week (and this despite the fact that almost all three-dimensional packages did not start or immediately crashed). Therefore, the comparison with Intel Atom was carried out only in synthetic tests, alas.

At the same time, a very interesting system based on a dual-core processor from the CULV line, SU4100, was found in the bins of the test results. Despite the fact that the processor is considered outdated, at one time it was created as an inexpensive energy-efficient solution, that is, in positioning it is close to the older version of AMD Brazos. Therefore, it was decided to include it on the list. But we did not include systems based on Core i5 and older Core i3 in this comparison, this is a completely different class of processors. They are more productive, but they also consume more energy. For comparison, we took the weakest Core i3-350M tested to see how much faster it is. Some individual tests mention other systems as well.

Notebook nameAMD Aspire One AO522eMachines E644Acer aspire oneDell Inspiron 1470ASUS K42j
CPUAMD C-50AMD E-350Intel Atom N450Intel SU4100Intel Core i3-350M
Number of Cores2 2 1 (2 streams)2 (2) 2 (4)
Rated frequency1000 MHz1600 MHz1.66 GHz1.3 GHz2.26 GHz
Voltage1.05-1.35V1.25-1.35V0.8-1.1175Vn / an / a
Max power consumption9 watts18 watts6.5 W *10 W *35 W *
Video subsystemRadeon 6250Radeon 6310NM10N / aIntel HD Video

* There is some confusion in the power consumption data, as AMD usually indicates maximum power consumption, while Intel usually indicates the typical one, which is less. Therefore, the comparative data in this column should be viewed critically.

Intel's lineup has two closely related processors, the N450 and N455. They are no different, except for support for DDR3 in the second case, the N455 was released a quarter later and for some reason it has 1 W more thermal package despite the fact that all other characteristics and even the price are the same. It is possible to compare processors, when comparing it can be seen that although Intel, it would seem, provides comprehensive information, there are still many "gray areas" in the characteristics.

Atom is manufactured using 45nm technology, while AMD processors are manufactured using 40nm technology. But the Atom has a lower supply voltage, that is, in theory, it should be more economical ... But how will the platform and graphics behave?

Comparison in synthetic tests

First, let's make a rough comparison in synthetic tests. To do this, we use a traditional set of packages, the first of which are two tests. Cinebench 10 and 11.5 ... Of the synthetic tests, I trust this more, since it is still built on a real engine.

Cinebench 10.0Cinebench 11.5
1 CPUAll CPUOpenglOpenglCpu
AMD C-50665 1266 1419 5,07 0,40
AMD E-3501062 2048 2037 7,72 0,64
Intel Atom N450566 866 289 ---* 0,27
Intel Atom D525622 1714 323 (1278) 6,18 0.56
Intel SU41001561 3030 668

* The test fails because the video core does not support the required functions.

What conclusion can be drawn from the test of the 10th version? Targeting netbooks and tablets, the weaker C-50 outperforms the Atom N450, with which it is approximately equal in power consumption and lags slightly behind the Atom D525, but this model is much more power hungry, even Intel specifies a 13W thermal package for it. So for its niche even CPU performance is not bad. The E-350 is faster than the Atom line of processors, but lags well behind the SU4100.

We should also pay special attention to the performance in OpenGL. Intel's integrated video is very weak and cannot compete with AMD products. The NVIDIA ION2 result (the result in brackets for the Atom D525, this platform was used in the ASUS EEE PC 1215N netbook) can already compete with the younger AMD Brazos model (although it lags behind the older one). But the economic feasibility of building such a platform is a big question, because it is a complete Intel Atom platform (processor plus a chipset), on which another graphics chip hangs through an external interface. An expensive, uncomfortable design, crafted out of desperation. Yes, and ION2 was inserted there, apparently, to get support for the HDMI interface.

In Cinebench 11.5, the alignment of forces in OpenGL has changed slightly - now the NVIDIA adapter takes the middle position between AMD solutions. However, all three had poor results. By the way, interestingly, the situation in processor tests is about the same.

In general, the 525th model from Intel is between two AMD platforms in terms of performance (and should be far behind in terms of consumption, since its thermal package is almost twice as different from Atmo N450).

Let's look at the test PCMark Vantage.

PCMark VantageAMD C-50AMD E-350Intel Atom N450Intel Atom D525
PCMark Score1520 2132 1286 1832
Memories score1244 1653 430 1550
TV and Movies Scorefailfailfail741
Gaming Score1400 1877 580 1826
Music score1492 2541 1885 2431
Communications Score1548 2318 1167 1551
Productivity Score1228 1413 1085 1804
HDD Score2462 2714 2688 3156

In the jungle of PCMark results, I provide readers with an opportunity to understand. Although the final score is in general roughly the same as the results of Cinebench. It is difficult to comment on the results of the subtests, so we will not do this, but move on to testing in real applications.

Testing in real applications

Real-world testing follows the 2010 test procedure. Let me remind you that the results of specific applications can be compared for all mobile and desktop systems (except for games, the settings in this group were seriously changed, and the parameters of the test problem for Photoshop, where the size of the test file was reduced). But this applies only to the test results themselves, it is impossible to compare the rating figures, since they are calculated based on different sets of applications.

If there are blank columns in the table, this means that either the test did not work out correctly, or it is impossible to calculate the rating correctly.

Let's start with professional apps.

3D visualization

This group contains applications that are demanding on the performance of the processor and graphics subsystem. Therefore, the results of their work are of purely academic interest.

AMD E-350Intel SU4100
Lightwave - work67.25 172.38
Solidworks - work94.8 334.13
Lightwave - rating37 15
Solidworks - rating71 20
Group - rating54 18

Only two systems passed the full test, the E-350 and the SU4100. The weak C-50 predictably didn’t pull, the Lightwave test didn’t work in the i3-350M, so its results were excluded from consideration. This is the first win for AMD in this group. And in both applications.

3D rendering

Let's see how things are in the rendering of the final scene, where the main load falls on the central processor. There are still only two participants.

AMD E-350Intel SU4100
Lightwave665,02 633,93
3ds max0:48:44 0:40:28
Lightwave - rating20 21
3ds max - rating23 28
Group - rating22 25

And here the AMD processor is slower. True, it must be said that both processors performed the test for a very long time; it is definitely not worth using them in such applications in real life.

Calculations

This group measures the mathematical performance of the processor. Let's see…

AMD C-50AMD E-350Intel SU4100Intel Core i3-350M
Solidworks 128,93 101,69 53,99
MATLAB0,2846 0,1859 0,1192 0,0651
Solidworks - rating 40 51 96
MATLAB - ranking20 30 47 86
Group - rating 35 49 91

AMD processors don't look so good anymore. The E-350 turns out to be weaker than the SU4100. But this is already a rather old processor, moreover, it is also focused on energy efficiency, and not on performance.

Compilation

Test for the compilation speed of the program using the Microsoft Visual Studio 2008 compiler.

AMD C-50AMD E-350Intel SU4100Intel Core i3-350MAMD E-350 Desktop
Compile0:46:06 0:30:38 0:22:07 0:09:26 00:30:49
Compile - rating17 26 36 85 26

Firstly, for this benchmark there are results for the E-350 processor in the desktop system, and we can see that the results are practically the same - both in the laptop and in the desktop board.

Let's look at the balance of power. The S-50 is in the deep tail of any comparison. Such low results make one think: the processor may be too weak even for some home tasks, such as flash video.

E350 lost even CULV in both variants and lags far behind Core i3.

Java Application Performance

This benchmark represents the execution speed of a set of Java applications. The test is critical to the speed of the processor and reacts very positively to additional cores.

AMD C-50AMD E-350Intel SU4100Intel Core i3-350MAMD E-350 Desktop
Java12,62 19,92 24,8 56,73 21,87
Java - ranking14 22 28 64 25

Interestingly, the balance of power in this test remains practically the same. There is a noticeable difference between the desktop and mobile versions of the E-350, the desktop version has gone ahead. Because of which? Faster memory?

Both AMD processors lag behind Intel's solutions, but they will almost certainly be noticeably faster than the Atom.

Let's move on to productive home tasks: working with video, sound and photos.

2D graphics

Let me remind you that there are only two tests left in this group, which are quite diverse. ACDSee converts a set of RAW photos to JPEG. And Photoshop performs a series of image processing operations - applying filters, etc. The Photoshop test results cannot be directly compared, since the test file is reduced (this is done to make the test work better on systems with a small amount of RAM).

AMD C-50AMD E-350Intel SU4100Intel Core i3-350MAMD E-350 Desktop
ACDSee0:21:26 0:14:57 0:10:22 0:06:43 00:13:59
Photoshop0:11:44 0:04:09 0:03:07 0:01:38 00:17:59
ACDSee - rating35 51 73 113 54
Photoshop - rating47 132 175 335
Group - rating41 92 124 224

In the ACDSee test, there is again a noticeable difference between the E-350 processor in a laptop and a desktop.

Whatever one may say, but the noted alignment of forces remains here as well. We can make a preliminary conclusion that in situations where only the performance of the processor is needed, the AMD E-350 even outperforms the relatively old Intel SU4100.

Audio encoding in various formats

Encoding audio into various audio formats is a fairly simple task for modern processors. The dBPowerAmp wrapper is used for encoding. She knows how to use multicore (additional coding streams are launched). The test result is its own points, they are inverse to the time spent on coding, that is, the more, the better the result.

AMD C-50AMD E-350Intel SU4100Intel Core i3-350MAMD E-350 Desktop
apple26 40 47 104 41
flac30 49 61 138 49
monkey23 36 45 101 37
mp313 21 26 62 22
nero12 19 24 59 19
ogg8 13 18 43 14
apple - rating16 24 29 63 25
flac - rating15 24 30 69 24
monkey - rating16 24 31 69 25
mp3 - rating15 24 30 72 26
nero - rating15 23 29 72 23
ogg - rating14 22 31 74 24
Group - rating15 24 30 70 25

The test is quite simple, but at the same time, illustrative. On the whole, he confirms the noted tendency.

Video encoding

Three out of four tests are encoding a video into a specific video format. The Premiere test stands apart, in this application the script provides for the creation of a movie, including the imposition of effects.

AMD C-50AMD E-350Intel SU4100Intel Core i3-350MAMD E-350 Desktop
DivX1:00:42 0:12:31 0:09:41 0:05:23 00:12:21
Premiere0:52:26 0:29:55 0:20:12 0:07:28 00:29:24
x2641:35:48 0:56:04 0:36:56 00:57:28
XviD0:59:01 0:09:37 0:07:23 0:04:12 00:09:18
DivX - rating7 35 45 80 35
Premiere - rating10 17 25 68 17
x264 - rating11 19 28 18
XviD - rating5 32 42 73 33
Group - rating8 26 35 26

Immediately striking is the catastrophic lag of the S-50. The rest of the processors are keeping in line with the already noted tendency: the E-350 lags behind the SU4100, the i350M is far ahead.

Finally, there are several types of household tasks.

Archiving

Archiving is a fairly simple mathematical problem in which all processor components are actively working and the final performance depends on all components.

AMD C-50AMD E-350Intel SU4100Intel Core i3-350MAMD E-350 Desktop
7-zip0:13:26 0:08:54 0:06:51 0:03:16 00:08:39
WinRAR0:07:44 0:05:13 0:03:45 0:02:33 00:05:12
Unpack (RAR)0:03:23 0:02:16 0:01:41 0:01:10 00:02:16
7-zip rating17 25 33 68 26
WinRAR - rating32 48 66 97 48
Unpack (RAR) - rating34 51 69 100 51
Group - rating28 41 56 88 42

One of the most obvious and simplest tests. The results are quite illustrative, they can be used to assess the level of processor performance.

Performance in browser tests

Quite simple tests too. Both measure performance in Javascript, which is perhaps the most performance-intensive part of the browser engine. The trick is that the V8 benchmark scores in points, while the Sunspider scores in milliseconds. Accordingly, in the first case, the higher the number, the better, in the second - vice versa.

AMD C-50AMD E-350Intel SU4100Intel Core i3-350MAMD E-350 Desktop
Googlev8-chrome1517 2419 3023 2137 1622
Googlev8-firefox118 202 255 190 203
Googlev8-ie44 52 66 51 54
Googlev8-opera899 1391 1689 1265 1409
Googlev8-safari595 933 1165 920 942
Sunspider-firefox3138 2015 1662 2155 2002
Sunspider-ie17928 11323 9078 13497 11133
Sunspider-opera1185 758 698 897 801
Sunspider-safari1751 1146 915 1210 1362
Googlev8 - ranking34 51 64 48 48
Sunspider - rating37 57 69 52 55
Group - rating36 54 67 50 52

The results of this test roughly correspond to the noted trend, except for a strange dip in the results of the i350M, most likely caused by technical reasons.

Comparison in HD Play

This test has been removed from the desktop benchmark, although it is still valid on mobile. Even if the system copes with decoding, in a laptop it is very important how much resources it requires for this. This is both heating and battery life ...

AMD C-50AMD E-350Intel SU4100
H.264 hardware41,1 27,5 20,7
H.264 software76,5 81,2 78,9
H.264 hardware - rating40 60 79
H.264 software - rating44 42 43
Overall rating42 51 61

Let's take a closer look at this test, because almost every user can face it.

Basically, both AMD processors can handle even 1080p software decoding. Although, I would say that it is "on the edge": almost always with such a high processor load, the system already starts to skip frames and playback loses smoothness. In the program mode, the load for all processors is about the same, for some reason the lowest is for the weakest C-50.

When hardware acceleration was enabled, the places were immediately allocated in the usual way, although I thought that AMD systems would be ahead here because of the good optimization algorithms in ATI video chips. However, this did not happen.

Well, time to jump to conclusions.

Overall system rating

Let's look at the average score of the systems that participated in the test.

AMD C-50AMD E-350Intel SU4100
E350 vs SU4100 comparison 40 47
Comparison of the three systems22 39 53

In the first row, the ratings are calculated for two systems (based on the ratings of all applications launched on them), i.e. AMD E-350 and Intel SU4100, in the second - for three systems, only applications that have started and worked on all three are taken into account. systems.

Let's briefly summarize the impressions of the performance tests. It is immediately striking that the Atom was not tested in real applications, but the SU4100, which was already leaving the scene, took part. At the same time, in desktop testing, where the E-350 and Atom were compared, both processors could not come close to the outdated and cheap Celeron processors. I’m afraid that they would not make the same claims to this article - they say, where did I find the laptop with the SU4100? Indeed, Intel now hardly promotes this line (and in vain), I think that soon it will disappear altogether from everywhere, if it has not disappeared yet. And why there are no Atom results.

In the near future we will definitely try to measure the performance of a netbook based on Intel Atom and publish comparative results. However, based on the results of synthetic tests, I would assume that the processors from the mobile line will be weaker than the C-50. Moreover (although it is not a fact that the tests will show this), due to the more powerful graphics subsystem, AMD processors should be more comfortable in everyday work. E-350 in netbooks should become a leader in speed.

Although the catch is that the E-350 is positioned in more productive segments than netbooks. And the picture is interesting: Intel had a processor for the same niche, also energy efficient and not very slow. It was also installed in large 15-inch laptops, citing the fact that performance is not so important in office systems. The processor, by the way, turned out to be not very popular and is now leaving the scene. And AMD seems to be trying to play in this segment again. On the other hand, now in Intel products there should be a gap between too slow Atom and more productive, but also gluttonous modern Core. The AMD E-350 falls into this gap and looks good for a certain product category, if the SU4100 didn't seem to be more interesting.

Power consumption and battery life

Let's see how much power a laptop with a particular processor consumes in different scenarios. Unfortunately, we do not have data on the SU4100 and Core i3-350M (these notebooks were tested before we even measured the power consumption). But we can measure the power consumption of a system based on Intel Atom, its results should certainly be more interesting than that of CULV.

* 18W @ minimum backlight brightness.

** 27W no HDD load.

The results were a bit unexpected for me. It turns out that the new system on the C-50 consumes slightly more power than the system on the Atom N450 (for measuring the system on the Atom 450, we used the MSI Wind 160 model with standard power consumption drivers from Microsoft). Of course, we are talking about the power consumption of the system as a whole (including the screen, etc.), but the systems are very close, two netbooks with practically the same characteristics. Since the power consumption is close, then the autonomy of devices with an AMD C-50 APU should also be about the same as that of devices based on Atom, and for tablets, for example, this is not a very good option.

However, given the same power consumption, the C-50 is faster than the N450, and the D525 certainly consumes more power and, more importantly, dissipates much more heat. By the way, the C-50 netbook is much colder than its Atom competitor.

E-350 is not a leader in terms of energy consumption either; in economical modes it is close to portable models based on Intel. Although, if we draw analogies again, its energy consumption should be close to the CULV systems, and at one time they performed very well in terms of autonomy: laptops with them easily pulled for 7-8-10 hours.

We will provide more detailed data on energy saving and other information in the reviews dedicated to specific laptops on the AMD C-50 and E-350 platforms.

conclusions

Once again, you take new AMD products in your hands with the thought that they will change the world, and you give them away with the thought “just one more processor, somewhere a little better, somewhere a little worse”. This, by the way, is not very good for the product, because too high expectations lead to disappointment in real research, and disappointment forms the wrong impression about the product, not allowing to correctly assess its advantages. However, AMD's new processors are a step forward. Let's try to estimate which one?

Firstly, the AMD Brazos platform surpasses Intel Atom in performance even of the processor core. Atom mobile series can only compete with the younger version, which operates at a much lower clock speed, and with the parity of frequencies, the AMD platform is far ahead. Moreover, what is important, this difference in performance is manifested in the segment where it is very important (since, let's face it, the overall performance level of all products is very low).

However, there is a subtle point about positioning here. While Brazos are faster than Intel Atom, they are generally in the same segment. On the one hand, this is good, since products based on them can easily integrate into the existing positioning systems of manufacturers' products. On the other hand, within the framework of this positioning, they can get the label “a cheaper alternative to Intel Atom”, with which they will remain, content with low profit and those users who do not want to pay for the Intel platform at all.

The second important advantage of the AMD Brazos platform: much more powerful graphics, both in performance and functionality. Now that graphics are used to speed up rendering, even in Internet browsers, a powerful graphics chip will definitely not be superfluous. Moreover, the main processor does not shine with high speed, so help will be very useful to it. In terms of performance, AMD Brazos is much more successful with HD content, which is important for a platform of this class. In terms of functionality, a modern graphics core with support for DirectX 11 is used, and also immediately and fully supports HDMI digital video output. This is a significant advantage for final products - both motherboards and mobile devices, netbooks and tablets. Only these advantages still need to be conveyed to the user, to convince him that they are important, and this is already the task of building the correct marketing policy on the part of AMD and manufacturers. Let's hope for the best, although previous experience in this area gives rise to some concerns.

From a constructive point of view, the most important advantage of an APU is that it is one chip, so that the platform is much cheaper, more compact, and colder than the competitor's multi-chip solutions. However, this advantage is greater for developers and manufacturers. What does it actually matter to a user how many chips are in his device? He needs it to be productive, cold and with good autonomy. And by what means this was achieved - the second question. Moreover, it is not a fact that if the manufacturer saves on production, then the final products will become cheaper.

But the lack of heating is an important advantage for the user. In my opinion, the heating level of both platform options is extremely low. The Acer 522 netbook with the C-50 processor still managed to warm up quite a bit, but the heating even in the hottest place reached 31–32 degrees Celsius, and the air blown out was not very warm either. And the model Aspire One Happy on the Atom 450 from the same manufacturer was just hot, it was unpleasant to hold the netbook on your lap. But the processor there is the "coldest" of the Intel line. E-350 is not able to heat the laptop case at all. The eMachines 644, which we tested, stayed cold at all times, only the hard drive warmed up (and warmed up the case a little). At the same time, the cooling system in both laptops worked almost silently.

Summing up, we can say that the products themselves turned out to be very good. In terms of speed, functionality and other parameters, they are well suited for the tablet and netbook market, which is now actively developing and evolving, and can serve as a basis for very interesting devices. It's a shame that they came out too late, if they hit the market in 2010, during the netbook boom, a powerful, versatile, low-light platform with HDMI support could have made a splash.

However, even now the time has not yet been lost. But you shouldn't let things go by themselves. In order for AMD Brazos to be successful in the market, they need to be actively promoted among both consumers and manufacturers. And here there are some concerns. Because, firstly, AMD has an image of a manufacturer of "inexpensive alternative to Intel products", which forces buyers and manufacturers to discard interesting functionality and focus only on price, which is fundamentally wrong. Very often, the fact that a product is built on the AMD platform means that it is functionally poorer, lacks additional features, a good bundle, etc.

For example, Brazos can be a great base for HTPC, but one platform is not enough for this. On its basis, it is required to build an interesting final product with the necessary functionality and (this is important!) A good set of delivery. Who will make them and how do you bring them to market?

The same applies to the segment of laptops and netbooks. Potentially, the platform can be very successful if it is presented correctly (emphasizing the existing significant advantages) and if the initiative is supported by manufacturers, releasing really interesting solutions, and not just another super-cheap model from the “to be” series (which can ruin the most interesting technological solutions). And it is not worth, especially at the present stage, to get involved in dubious adventures such as organizing new incomprehensible market niches (what are they going to do with the E-350), and even more so not to try to position the processor against competitors, to whom it loses in terms of performance. Incidentally, Intel's CULV platform is a good example of marketing failure. For some reason, a weak, but economical processor began to be shoved into 15-inch desktop laptops, which killed the whole idea. Let's hope AMD doesn't repeat this mistake.

Summing up, I would like to say that AMD Brazos is a necessary and interesting product for its segment. But its success will largely depend not on the technological and technical advantages of the platform, but on the correct marketing and competent promotion of the platform to the market. Only in this case the platform will be successful. Otherwise, it will remain another uncommon niche solution, examples of which we have already seen on the market.