Socket am2 which processors it supports. "Socket AM2" from AMD: models of processors, sets of system logic and hardware platform specifications. Final rendering of 3D scenes

To the processor socket Socket AM2. Then we noticed a slight increase in performance where it was, and a change in the rating system. Today we continue our excursion to Socket AM2 and see what it gave to ordinary (single-core) AMD Athlon 64 processors.

AMD Athlon 64 AM2

Let's remind that the transition to Socket AM2 was necessary in order to give AMD processors the ability to work with faster DDR2 memory, thereby increasing the system performance based on them. Unlike the budget Sempron line, Athlon 64 processors received support not only for DDR2-400 / 533/667, but also for DDR2-800. Otherwise, no other significant changes have occurred, either architecturally or in the rating systems. Let us recall the main characteristics of new and outgoing processors, in the form of tables: Athlon 64 Socket AM2

	CPU frequency, GHz	HT frequency, MHz		Technical process

Athlon 64 Socket 939

	CPU frequency, GHz	HT frequency, MHz		Technical process	Dual channel memory controller

Athlon 64 Socket 754

	CPU frequency, GHz	HT frequency, MHz		Technical process	Dual channel memory controller
				90nm / 130nm, SOI
				90nm / 130nm, SOI
				90nm / 130nm, SOI
				90nm / 130nm, SOI
				90nm / 130nm, SOI
				90nm / 130nm, SOI

As you can see from the tables, the memory subsystem acceleration did not affect the rating system. But the lineup has decreased. This is partly due to the rejection of the production of more expensive chips with 1 MB L2 cache, which were quite good competitors to Athlon 64 X2, especially in games. In addition, at the beginning of next year, there are tendencies of displacing the entire line of Athlon 64 processors with dual-core X2, the price of the younger models of which (Athlon 64 X2 3600+) should approach the $ 100 mark by the end of this year, while Sempron processors should also become dual-core and supplant Athlon 64 from the bottom. But we will not bury yet, quite new, processors.

If we compare the sizes of the boxes, the packaging for AM2 has become more compact, which can be positively characterized - it will be more convenient to take away many processors.

Inside the package you will find: a processor, an "updated" cooler, a user manual and a logo sticker - nothing unexpected.

AMD Athlon 64 Socket 939 and Socket AM2 on top

As already noted, the updated processors have very few external changes. Above they are given only by the marking, which now looks like ADA3200IAA4CN. Everything is roughly deciphered as follows: ADA - Athlon 64 for workstations, 3200 - processor rating, I - 940 pin OµPGA package type (Socket AM2), A - alternating core voltage (≈1.25-1.35 V), A - variable maximum allowable temperature (≈65-69 ° C), 4 - 512 KB L2 cache size, CN - Orleans core.

AMD Athlon 64 Socket 939 and Socket AM2 bottom

On the bottom, the processor for Socket AM2 is relatively easy to distinguish by the extra leg (in the photo it can be found on the right processor in the lower left corner). And now a complete informational summary of the tested processor and used GEIL DDR2-800 memory obtained using the CPU-Z utility.

For comparison, we present information about AMD Athlon 64 3200+ Socket 939 with DDR-400 Hynix.

Overclocking

The test sample Athlon 64 3200+ with a standard boxed cooler was overclocked to 2700 MHz almost on the fly, but further increase in frequency resulted in a decrease in the system stability.

At the same time, the GEIL DDR2-800 modules were able to run in DDR2-900 mode, albeit with an increase in Command Rate to 2T.

Testing

To compare the performance of Socket 939 and Socket AM2 platforms, the following test systems were assembled, differing, apart from processors, in motherboards and RAM. Testbed for Socket 939: Testbed for Socket AM2:

Before directly comparing Athlon 64 Socket 939 and Socket AM2, we decided to investigate how sensitive the latter are to the speed of RAM. To do this, using the BIOS settings, we turned DDR2-800 into DDR2-667, DDR2-533 and DDR2-400 (timings were set according to SPD) and checked how the performance changes.

GEIL DDR2-800 in DDR2-667 mode

GEIL DDR2-800 in DDR2-533 mode

Since the processor core has not undergone changes, the performance does not change much, even with a significant acceleration of the RAM. So on Socket AM2, judging by the results of synthetic tests, a small increase in performance can be observed only in resource-intensive applications that are demanding, first of all, to the volume and performance of the memory subsystem, the increased clock frequencies of which are eaten up by increased latency and, possibly, some shortcomings in controllers memory. Let's move from synthetics to practice:

A surprise was received immediately, in Quake 3, which turned out to be very sensitive to memory latency and revealed an imperfect memory controller. The test became a smooth transition from synthetic tests to the results obtained in modern games.

The Socket AM2 platform disappointed a little with the drop in performance in games - although the result is not much worse, and in some places the same, but, unfortunately, not better, which we very much expected.

conclusions

As our testing has shown, having acquired support for faster DDR2 memory, AMD Athlon 64 processors not only did not improve, but also slightly lost in performance in most tasks. Accordingly, there is no point in recommending "transferring" to a new platform. But when assembling a new system, you will have to think about it and answer the question for yourself: "Is this the final configuration of the system or do I plan to upgrade after a while?" If after a while there is a desire to replace the processor, say, with a dual-core processor, and increase the memory, the Socket AM2 system will look much more promising - it will be not only cheaper, but also easier to upgrade. In addition, Socket AM2 has already provided a small performance boost in some tasks - if they are basic, then you will have to think even less. We express our gratitude to the company PF Service LLC (Dnepropetrovsk) for the processors and other equipment provided for testing.

Introduction The latest financial reports released by AMD show that the company is shipping fewer and fewer desktop processors every quarter. I must say that this trend should not cause any surprise, at least among our readers. Unfortunately, AMD's processor architectures are developing in such a way that the processors it produces are becoming less and less interesting for desktop users, and even more for enthusiasts.

You don't have to go far for examples. The flagship AMD FX series stopped developing for a long time, and the processors offered in its composition today not only lose in all consumer characteristics to the competitor's CPUs, but also have noticeably outdated characteristics. The middle class - hybrid processors - are oriented more towards mobile applications, and their desktop incarnations, although they are periodically updated, remain niche products with not too wide scope of applicability. In addition, quite unpleasant things sometimes happen to them: for example, the recently released APUs of the Kaveri family, focused on use in desktop systems, turned out to be slower than their predecessors, which, of course, does not add to their attractiveness. Naturally, in such a situation, even the most loyal fans of this company gradually turn away from AMD products.

At the same time, the manufacturer does not give any hope for a quick change in the current situation. AMD's current plans for new high-performance CPUs do not promise in the near future, and future APUs will certainly continue to move on the path of priority optimization of power consumption, but not performance. However, AMD has not yet lost all its baggage, potentially applicable to desktop processors. In addition to the Bulldozer microarchitecture branch, which has now evolved to the Steamroller version, the company also has another microarchitecture in its arsenal - Bobcat, which later grew into Jaguar.

While the development of Bulldozer followed the path of optimizing power consumption and reducing the performance of processors built on its basis, the original energy-efficient Bobcat-Jaguar microarchitecture moved in the opposite direction - towards increasing performance. And along the way, AMD has achieved some success. Originally aimed at low-cost and undemanding computers like netbooks and nettops, Jaguar's microarchitecture has been able to infiltrate higher-end devices - gaming consoles. This victory became an important milestone for AMD: the company secured itself with orders for several years ahead and created a kind of aura of a successful CPU developer around itself. And now, inspired by the success, she wants to try to get Jaguar recognition in the desktop market.

Kabini processors, built on the Jaguar microarchitecture, have long been used in mobile computers. Therefore, from AMD's point of view, they may well be in demand in the increasingly popular compact form factor desktop systems, if, of course, they can offer characteristics comparable to competing options. And in order to give its newly-minted Jaguar incarnations the status of full-fledged desktop processors, AMD has developed a new Socket AM1 ecosystem for them, and has also prepared a whole line of corresponding models.

The manufacturer claims that due to its low cost, this platform will be able to make a splash in the field of entry-level systems, which are especially in demand in emerging markets. For example, within the presentation of Socket AM1, a strong emphasis was placed on the countries of Latin America: it is there, according to AMD, that desktop processors based on Jaguar are simply doomed to success.

However, in fact, Kabini is not so hot what a novelty. Such processors have been available on the market for almost a year, and no one has interfered with their introduction into desktop PCs before. However, there were not many who wanted to contact them. The reason for their low popularity was that building desktop systems based on Kabini until recently required manufacturers to independently develop the design of motherboards, and the demand for such solutions was not clear. But now the situation has changed. Processors based on the Jaguar microarchitecture, in the wake of the start of sales of game consoles, arouse interest among consumers, and AMD is ready not only to work closely with manufacturers on the development of motherboards, but also to invest in promoting the Socket AM1 platform. As a result, Socket AM1 motherboards and processors will soon become widely available on store shelves, where they will delight the eye with their intriguingly low price. Whether the buyers who follow this bait will later regret their purchase, we will try to understand by testing the new Kabini in common tasks.

Desktop Kabini Architecture Details

The announcement of the Kabini socket-installable for low-budget systems is a game changer in this market. Until now, such processors, including Intel's Atom or AMD Zacate, were usually soldered onto motherboards. However, AMD considered that the availability of CPU upgrades could be one of the key factors in the market for budget energy efficient platforms, and decided to introduce replaceable CPUs. There is a certain logic in such a decision: the possibility of an upgrade is something that can attract buyers who previously preferred inexpensive tablets, netbooks, nettops, chromebooks and similar surrogates of full-fledged personal computers.

At the first stage, four processor options are offered for use in the Socket AM1 platform:

All of these processors are based on semiconductor crystals, manufactured using 28nm technology, and consist of four or two computing cores with Jaguar microarchitecture and a graphics core with modern GCN architecture with 128 shader processors. That is, Kabini offered in the version for the Socket AM1 platform are very similar in characteristics to similar mobile processors that have been available for almost a year. Athlon 5350 is similar to A6-5200, Athlon 5150 is closely analogous to A4-5100, and Sempron 3850 and Sempron 2650 processors are close relatives of E2-3800 and E1-2500. There is a slight difference only in the frequencies of the graphics core and in the TDP rates, but in general, the new desktop Kabini are no different from the old, mobile ones. And this is actually quite sad: over the past year, AMD has not been able to do anything with the frequency potential of its junior CPU line.

Those users who thought that the Socket AM1 platform would allow themselves to create something similar to the latest generation of SONY or Microsoft game consoles will also remain upset. The processors used there each have 8 Jaguar computing cores, operating at a frequency of just below 2 GHz, and a graphics core with GCN architecture, which has no less than 768 shaders. In other words, the new desktop Kabini is very, very far from console APUs.

Obviously, AMD is focused on the lower price segment, and presents the Socket AM1 platform as a further development of the Brazos 2.0 platform. If we compare Kabini with Zacate processors, then they are, indeed, noticeably more advanced offerings. If only because the number of processing cores has doubled in new CPUs.

Noticeable changes have been made to the Jaguar microarchitecture itself, which contains certain improvements over the previous Bobcat microarchitecture. However, they, like in the Bulldozer branch, are not of a fundamental nature. Jaguar's energy-efficient microarchitecture remains designed to execute only two instructions per clock cycle, making it similar to Intel's Silvermont microarchitecture found in the Bay Trail series. Naturally, as before, Jaguar uses out-of-order command execution. Still, the main changes in this microarchitecture are aimed at improving the efficiency of the resources available since the Bobcat times, and therefore are concentrated in the input part of the executive pipeline.

First, an additional 128-byte loop buffer has been added to the L1 instruction cache. It allows you not to engage in multiple fetching of instructions from the L1 cache in cycles, but in fact, this does not increase performance, since its latency is no less. The point of this improvement is solely to reduce consumption. Second, at Jaguar, AMD has improved how the instruction prefetching mechanism works. Third, in the new microarchitecture, the size of the buffer between the L1 cache and the instruction decoder has been increased, which has made it possible to somewhat reduce the dependence of the instruction fetching and decoding processes. And fourthly, the execution pipeline has been extended by one stage related to the decoding stage. The purpose of this change is to improve the frequency potential of the new microarchitecture, which in Bobcat was limited precisely by the poorly designed decoder.

There are changes at the stage of command execution. First of all, it should be noted that in Jaguar the command system is tightened to a more current state. Added SSE4.1 / 4.2, AES, CLMUL, MOVBE, AVX, F16C and BMI1 to the supported instructions. Such innovations required a redesign of the floating point block. While the FPU in Bobcat was 64-bit, in Jaguar this unit was completely 128-bit. As a result, 256-bit AVX instructions are executed in two steps, but 128-bit instructions no longer require any division into parts. At the same time, the pipeline for processing real-valued operations in Jaguar lengthened by one stage, but, nevertheless, the productivity of vector operations in the new microarchitecture should be significantly higher than that of its predecessor.

There are also changes in the execution of integer commands. While Bobcat's performance on regular code was pretty good already, Jaguar introduced a new block for integer divisions, taken from the K10.5 microarchitecture. This allowed us to approximately double the throughput of the divisions.

In addition, AMD has increased the capacity of the scheduler buffers, which contributes to more successful operation of out-of-order execution algorithms.

The data loading and unloading unit in the energy-efficient Bobcat and Jaguar microarchitectures uses the same principles of operation as the analogous unit of "large cores". That is, it is capable not only of prefetching, but also of reordering queries. In the latest generations of the Piledriver and Steamroller microarchitectures, AMD has improved their prefetching algorithms and they are now ported to Jaguar. All this entailed an approximately 15 percent increase in the speed of the new microarchitecture with data.

All of the improvements made at the microarchitecture level raise the specific efficiency of the Jaguar core over the Bobcat core by about 17 percent. And if we add to this a possible increase in clock frequencies and the number of cores, then AMD promises an advantage of Kabini processors over Zacate at the level of 2-4 times.

By the way, the change in the structure of the processor module also played a significant role in increasing the speed in multi-threaded tasks. If earlier each of the cores had its own L2 cache (which worked, by the way, at half the processor frequency), and communication between the cores was carried out using an external bus, then Jaguar uses a scheme with a shared shared L2 cache. A single quad-core Kabini processor module includes a common roomy full-speed L2 cache up to 2 MB with 16-channel associativity. Moreover, for the first time for AMD, this cache has an inclusive architecture, that is, it duplicates data stored in the L1 cache. This requires an increase in the capacity of the cache, but it plays a positive role in combined multi-core work.

In general, thanks to the use of a more modern 28nm process technology and some computer-aided design techniques borrowed from the field of GPUs, one Jaguar core was able to fit into an area of \u200b\u200b3.1 square meters. mm, while the 40nm Bobcat cores used 4.9 sq. mm of area. In other words, adding a capacious L2 cache will not cause the crystal to swell and increase its cost.

The graphics core of the Kabini processor, along with the older APUs from AMD, received the latest GCN architecture, identical to the flagship graphics cards. As a result, all modern programming interfaces are supported by Kabini graphics: DirectX 11.1, OpenGL 4.3 and OpenCL 1.2. However, in terms of GPU power, Kabini is significantly curtailed. It is based on two computing clusters, that is, it contains only 128 shader processors, which is less than the lowest video cards of the Radeon R5 category. That is why the Kabini graphics core belongs to the Radeon R3 class. The 128 shader processors in the GPU come with eight texture units and four ROP units. In addition, the video core includes a command processor and four independent asynchronous computing engines responsible for distributing tasks under heterogeneous workload. However, HSA technologies are not supported in Kabini processors.

Despite the obvious weakness of the Kabini GPU processors, the VCE and UVD engines are fully preserved in it. This means that Kabini graphics can provide hardware support for video decoding in H.264, VC-1, MPEG-2, MVC, DivX and WMV formats, and in addition, it can hardware encode H.264 video content in FullHD resolution. However, for some reason, the latter option is not yet used in common utilities for transcoding.

Unfortunately, despite all the improvements in the architecture of the computational and graphics cores, the memory controller in Kabini remains single-channel. It supports DDR3-1600 to the maximum, so in many aspects of performance Socket AM1 systems may lack the memory bandwidth. Obviously, the already slow schedule will suffer first of all from this.

But the new desktop Kabini, like their mobile counterparts, are a full-fledged system-on-a-chip, in addition to computing cores, GPU, memory controller and north bridge, including the south bridge. It has a SATA 6 Gb / s controller, USB 3.0, as well as a PCI Express 2.0 controller that allows you to connect external devices to a Kabini-based system.

With the release of the socket-replaceable Kabini processors, AMD is reviving the Athlon and Sempron trademarks under which they will be sold. This may partly cause another confusion, since along the way AMD still supplies Athlon X4 processors for Socket FM2 with Richland design and Sempron 145 processor for Socket AM3 systems.

But the new Athlon and Sempron processors for low-cost desktop systems do push the price bar down a lot. The older version of the desktop Kabini costs only $ 55, and at the same time, the processor itself implements a full set of interfaces for creating a finished system. This means that the cost of Socket AM1 motherboards, which do not carry any expensive chips, can start from the $ 35 mark. Accordingly, the cheapest version of a desktop platform with a Kabini processor (requiring additions in the form of memory, storage and case) in this situation can cost as little as $ 65-70.

At such prices, there is nothing surprising: including 914 million transistors, the Kabini semiconductor crystal is very small - its area is only 105 square meters. mm.

AMD Kabini Semiconductor Crystal

AMD itself gives such an example: four Jaguar cores occupy about the same area on a die as a single dual-core Steamroller processor module.

Indeed, the core area of \u200b\u200bthe latest Kaveri processors is more than twice as large: it reaches 245 square meters. mm. Another analogy can be drawn: almost the same as that of Kabini, the core area has a dual-core Haswell with GT1 graphics (more specifically, it is 107 sq. Mm), for the production of which a more modern 22-nm technical process is used.

Socket AM1 platform

The new Socket AM1 platform, specially launched for cheap and energy efficient AMD processors, received its own processor socket, incompatible with anything other than the new Kabini itself, which until recently appeared in documents under the name Socket FS1b.

This processor socket in its design resembles "adult" AMD sockets, but it has a smaller number of contacts - 721 - and takes up a noticeably smaller area on the board.

To test the platform, we received an MSI AM1I motherboard made in Mini-ITX format. All motherboards for desktop Kabini will look like this.

It must be said that AMD wants to get Micro-ATX motherboards with Socket AM1 from manufacturers, but the most interesting at the price are compact motherboards with 17x17 cm format. For example, the MSI AM1I's recommended price is only $ 36. The reason for such a low price is clear from just one glance at the photo of the board. Socket AM1 processors allow making very simple motherboards. Even in the desktop version, Kabini remains a system-on-a-chip, which means that all the necessary controllers are integrated in it: DDR3 memory, PCI Express, USB and SATA buses. In other words, for the Socket AM1 motherboard to work, neither north nor south bridge is required, and the entire surface is allotted for placing small controllers and slots.

Kabini's built-in peripheral controllers provide support for:

Eight PCI Express 2.0 lanes, which can be routed to the PCI Express slot and to external controllers, for example, wired network, WiFi, etc .;
Two USB 3.0 ports and eight USB 2.0 ports;
Up to four 4K digital display outputs (DVI, HDMI, DisplayPort) and analog monitor output;
Two SATA 6 Gb / s channels without the ability to form RAID arrays;
SDXC UHS-I interface with a bandwidth of up to 104 MB / s for connecting SD cards.

Leveraging these capabilities, MSI has offered a motherboard equipped with two DDR3 DIMM slots that operate in single channel mode, a PCI Express x16 slot that is logically connected to four PCIe 2.0 lanes, and a mini-PCIe slot that can accommodate a half-format card. The board itself also has two SATA 6 Gb / s ports and two connectors for connecting four additional USB 2.0 ports. In addition, it is possible to connect serial and parallel ports, as well as a TPM module. The number of supported fans is limited to two, and the processor one is designed exclusively for a three-pin connection.

The rear panel of the board has two PS / 2 ports for a mouse and keyboard, D-Sub, DVI-D and HDMI monitor connectors, two USB 2.0 ports, two USB 3.0 ports, an RJ-45 socket for a gigabit network and three analog audio connectors. ... The Realtek RTL8111G controller is responsible for the built-in network operation, and the analog sound is output through the eight-channel Realtek ALC887 codec. It should be noted that the board can display images on two monitors simultaneously both in cloning mode and expanding the desktop. But monitors with a resolution above 1920x1200 only work with an HDMI connection.

The voltage converter for MSI AM1I is assembled according to a three-channel scheme, but for powering processors whose maximum consumption does not exceed 25 W, this should be quite enough. Moreover, the Socket AM1 platform does not provide for any overclocking. The maximum memory frequency that can be set through the BIOS is 1600 MHz, the processor multiplier does not change upward, and there are simply no settings for the base clock generator frequency.

Besides MSI, motherboards for Socket AM1 processors in Mini-ITX and Micro-ATX form factor have been announced by almost all brands. Note that until this moment there was no particular zeal among manufacturers in the release of motherboards based on economical AMD CPUs. Probably, in Socket AM1, Taiwanese marketers really saw some kind of perspective.

The new platform also introduces its own format for processor coolers, which have received a fundamentally new mount. While from time immemorial, on motherboards for AMD processors, coolers have been clinging to the teeth of the processor frame, the cooler for Kabini rests on two plastic dowels inserted into special holes in the PCB located on the diagonal passing through the socket. The distance between the mounting holes is small - only 85 mm.

The stock cooler itself is a relatively small aluminum heatsink, on which is mounted a buzzing fan with a 50 mm impeller diameter, a maximum speed of 3000 rpm and voltage control. Frankly speaking, it would be much more pleasant to see passive cooling in this case, but such a heatsink capable of dissipating up to 25 watts will not be cheap, which contradicts the ideology of the Socket AM1 platform. Nevertheless, a number of manufacturers of cooling systems still promise to support the new format, so it may soon be possible to purchase some alternative options in stores.

The release of Kabini in the form of processors installed in sockets primarily makes sense in the sense that it gives hope for the possibility of a subsequent upgrade of such systems. However, the prospects of Socket AM1 still remain a big question. On the one hand, AMD should switch from Kabini processor design to Beema, but AMD has not made any statements about the compatibility of these processors according to the conclusions. At the same time, it is quite possible that the desktop versions of Beema will have a DDR4 controller, which means that Socket AM1 platforms will become a dead-end branch, the modernization of which will not be feasible in practice. Besides, considering that the Kabini crystal also contains the south bridge, for compatibility purposes AMD should not add or change any interfaces in future Socket AM1 processors. In other words, if a manufacturer wants to add PCIe lanes, upgrade to a newer version of this specification, implement the ability to connect M.2 slots, or something similar, then most likely this will mean the need to upgrade to a new version of the processor socket.

Test processors: Athlon 5350 and Sempron 3850

To test the Socket AM1 platform, our laboratory received two models of such processors: Athlon 5350 and Sempron 3850.

AMD Athlon 5350



AMD Sempron 3850

In fact, they are similar to each other. And in both systems-on-a-chip there are four computing cores with the Jaguar microarchitecture, and the GCN graphics core has 128 shader processors. The size of the shared L2 cache in both cases is 2 MB. The belonging of these CPUs to different classes is determined by the clock frequencies.

The Athlon 5350 runs at 2050 MHz, while the Sempron 3850 runs at a much lower 1300 MHz.

AMD Athlon 5350



AMD Sempron 3850

The frequencies of the integrated graphics cores also differ. The older Athlon model has it at 600 MHz, while the Sempron 3850 has the graphics frequency reduced to 450 MHz.

The operating voltage of both processors is approximately 1.3 V, while in idle state, the frequency is reset to 800 MHz, and the supply voltage to 1.0375 V. The graphics core, without load, drops the frequency to 266 MHz. Kabini does not provide any options for turbo-mode neither for computing nor for graphic cores.

How we tested

Introducing its new Socket AM1 platform and the corresponding Kabini processors, AMD focused on the fact that these new items are positioned as an alternative to Intel's Bay Trail-D desktop processors: Celeron J1800, Celeron J1900 and Pentium J2900.

In the picture provided to us by AMD's marketing department, everything looks very good: Kabini processors are clearly more profitable in price.

However, the real situation is far from that shown in the illustration. First, the Bay Trail-D desktop Mini-ITX motherboards are actually noticeably cheaper, since Intel is releasing their systems-on-a-chip at significant discounts. For example, an ASRock or Gigabyte platform based on Celeron J1900 can be bought for about $ 80-90: that is, for about the same money as the Athlon 5350 bundled with the board. In this case, Intel's system will be much more economical. Typical heat dissipation for desktop Bay Trail-D modifications is set at 10 watts, and Kabini's TDP is two and a half times higher.

Secondly, among platforms based on Intel processors there is a variant more suitable for the role of competition with Socket AM1: desktop motherboards with integrated mobile low-voltage Celerons based on the Ivy Bridge microarchitecture. Mini-ITX motherboards built, for example, on the Celeron 1037U and similar CPUs, are available from Biostar, Gigabyte, Foxconn, Elitegroup and many other manufacturers. Their cost is approximately in the same range - about $ 70- $ 90, and the typical total heat dissipation of such processors together with the required chipset in this case is 21 W.

In other words, AMD opposes Socket AM1 with the Intel platform, which in fact is not its direct competitor. But we will not buy into this marketing ploy, so in our testing the Kabini desktop processors will be compared not only with the Bay Trail-D class Celeron, but also with the energy-efficient Celeron based on the Ivy Bridge microarchitecture.

In addition to Celeron J1900 and Celeron 1037U, we have included two "full-fledged" desktop processors of the lower price category among the competitors of Athlon 5350 and Sempron 3850: Celeron G1820 and A6-6400K. It should be borne in mind that they are not direct alternatives to Kabini, but their participation in the tests will allow us to draw conclusions about in what aspects the energy-efficient Socket AM1 platform is better or worse than the inexpensive Socket FM2 and LGA 1150 platforms, which can also be assembled on based on compact Mini-ITX motherboards.

As a result, the test systems were based on the following set of components:

Processors:

AMD A6-6400K (Richland, 2 cores, 3.9-4.1 GHz, 1 MB L2, Radeon R5);
AMD Athlon 5350 (Kabini, 4 cores, 2.05 GHz, 2 MB L2, Radeon R3);
AMD Sempron 3850 (Kabini, 4 cores, 1.3 GHz, 2 MB L2, Radeon R3);
Intel Celeron G1820 (Haswell, 2 cores, 2.7 GHz, 2x256 KB L2, 2 MB L3, HD Graphics);
Intel Celeron 1037U (Ivy Bridge, 2 cores, 1.8 GHz, 2x256 KB L2, 2 MB L3, HD Graphics);
Intel Celeron J1900 (Bay Trail-D, 4 cores, 2.0-2.41 GHz, 2 MB L2, HD Graphics).

Motherboards:

ASRock FM2A88X-ITX + (Socket FM2 +, AMD A88X);
Gigabyte C1037UN-EU (Celeron 1037U, Intel NM70);
Gigabyte J1900N-D3V (Celeron J1900 SoC);
MSI AM1I (Socket AM1 SoC);
MSI Z87I (LGA 1150, Intel Z87 Express).

Memory:

2 x 4 GB, DDR3-1866 SDRAM DIMM, 9-11-9-27 (Kingston KHX1866C9D3K2 / 8GX);
2 x 4 GB, DDR3L-1600 SDRAM SO-DIMM, 11-11-11-29 (2 x Crucial CT51264BF160BJ.C8FER).

Disk subsystem: Intel SSD 520 240 GB (SSDSC2CW240A3K5).
PSU: Corsair AX760i (80 Plus Platinum, 760W)
Operating system: Microsoft Windows 8.1 Enterprise x64;
Drivers:

AMD Chipset Drivers 14.4;
AMD Catalyst Display Driver 14.4;
Intel Chipset Driver 10.0.13.0;
Intel Graphics Driver 10.18.10.3498.

It should be noted that the memory in various test configurations was used at the maximum speed for each specific case. This means that AMD A6-6400K and Intel Celeron G1820 processors were tested with DDR3-1866, AMD Athlon 5350, AMD Sempron 3850 and Intel Celeron 1037U processors with DDR3-1600 memory, and Intel Celeron J1900 with DDR3- 1333 SDRAM.

Performance

Overall performance

To assess the performance of processors in common tasks, we traditionally use the Bapco SYSmark test suite, which simulates the user's work in real common modern office programs and applications for creating and processing digital content. The idea of \u200b\u200bthe test is very simple: it produces a single metric that characterizes the weighted average speed of a computer during everyday use. Recently this benchmark was updated once again, and now we use the latest version - SYSmark 2014.

The Kabini desktop processors, which are part of the Socket AM1 platform, occupy a place traditional for any AMD product on the diagram. During normal day-to-day use in common programs, their performance is noticeably lower than that of alternative options from Intel. This can be attributed to both the flaws of the Jaguar microarchitecture and the lack of "correct" optimization for AMD processors in popular software packages, but the fact remains. Even the fastest Socket AM1 processor Athlon 5350 lags behind the average Bay Trail-D model, Celeron J1900, by about 10 percent and is inferior to the energy-efficient dual-core Celeron 1037U by about 25 percent. In other words, the appearance of cheap desktop processors Kabini is unlikely to somehow change the usual market situation. Moreover, such AMD quad-core processors are many times behind the full-fledged budget Intel Haswell processors.

A deeper understanding of SYSmark 2014 results can provide insight into the performance scores obtained in various system use cases. The Office Productivity script simulates typical office work: preparing word, processing spreadsheets, working with e-mail, and surfing the Internet. The script uses the following set of applications: Adobe Acrobat XI Pro, Google Chrome, Microsoft Excel 2013, Microsoft OneNote 2013, Microsoft Outlook 2013, Microsoft PowerPoint 2013, Microsoft Word 2013, WinZip Pro 17.5.

The Media Creation scenario simulates the creation of a commercial using pre-shot digital images and video. The popular packages Adobe Photoshop CS6 Extended, Adobe Premiere Pro CS6 and Trimble SketchUp Pro 2013 are used for this purpose.

The Data / Financial Analysis scenario is devoted to statistical analysis and investment forecasting based on a certain financial model. The script uses large amounts of numerical data and two applications Microsoft Excel 2013 and WinZip Pro 17.5.

As you can see from the graphs, Socket AM1 systems do not shine with performance under any usage model. This means that in general they deliver lower performance than, for example, energy efficient and inexpensive competitor platforms. It is also quite curious that quad-core processors with Jaguar microarchitecture are inferior to all kinds of dual-core processors: both built on Ivy Bridge and Haswell microarchitectures, and on Piledriver. It turns out that due to the primitiveness of the internal design, the specific performance of Jaguar is very low, and increasing the number of simple cores still cannot be a good alternative in the x86 world to advanced internal processor algorithms.

In-app tests

We used the Cinebench R15 benchmark to measure the speed of photorealistic 3D rendering. Maxon recently updated its benchmark, and now it again allows you to evaluate the performance of various platforms when rendering in the latest versions of the animation package Cinema 4D.

It should be noted that when testing in Cinebench, the situation for Kabini processors is not so sad. The senior desktop representative of this family, Athlon 5350, is even ahead of its main competitors - Celeron J1900 and Celeron 1037U. This is natural. Jaguar's microarchitecture is well suited for executing parallelizable straight-line integer algorithms, which include the final rendering. However, the Sempron 3850 processor cannot share the success of its older brother - it sorely lacks the clock frequency to demonstrate acceptable performance.

Audio files transcoding speed testing is carried out using dBpoweramp Music Converter R14.4. This measures the speed at which converting FLAC files to MP3 format with maximum compression quality. The chart shows the performance in terms of the ratio of transcoding speed to playback speed.

This test is akin to the previous one. The Lame codec used here in multithreading works fine on Kabini processors. Athlon 5350 is even slightly ahead of the full-fledged dual-core Haswell, Celeron G1820. The reasons for Jaguar's good performance are the same - a non-branching algorithm based on integer operations.

We evaluated the high-definition video transcoding speed using the popular free utility Freemake Video Converter 4.1.1. It should be noted that this utility uses the FFmpeg library, that is, it ultimately relies on the x264 encoder, but it does some specific optimizations. During testing, we used the widely available DXVA technology to hardware accelerate the transcoding process.

Video transcoding is a more difficult task, but, nevertheless, Athlon 5350 pleases with good performance here too. It beats the Celeron J1900 of the Bay Trail family by 13 percent and the Celeron 1037U of the Ivy Bridge family by 27 percent. However, from the desktop Kabini, it seems that only senior representatives of the line can boast of good results in such tasks. The same Socket AM1 processors, which belong to the Sempron class, provide much lower and completely uncompetitive performance.

Considering that low-cost systems based on energy-efficient processors are often used as Internet terminals, special attention was paid to the performance of the Internet Explorer 11 web browser. Testing was carried out using the specialized Google Octane 2.0 Benchmark, which implements in JavaScript the ones actually used on the Internet -Application algorithms.

On the other hand, the internet performance of Kabini desktop processors is not very impressive. Yes, Athlon 5350 slightly outperforms the average Bay Trail-D model, Celeron J1900, but at the same time it lags far behind Celeron 1037U. But even this is not particularly upsetting, but how much the Socket AM1 platform turns out to be worse than "full-fledged" platforms during Internet activity. For example, even the dual-core Richland, A6-6400K, is exactly twice as fast as Athlon 5350.

We measure performance in the new Adobe Photoshop CC using our own benchmark, which is a creatively reworked Retouch Artists Photoshop Speed \u200b\u200bTest that includes typical processing of four 24-megapixel digital camera images.

The fact that Jaguar's microarchitecture would not shine in complex tasks like processing graphic images was immediately clear. However, in its justification, it should be emphasized that the energy-efficient Silvermont microarchitecture used in the Bay Trail is also not distinguished by high performance. In other words, processors built on "large" cores are more appropriate here, at least the same Celeron 1037U, which, like Kabini, has both low power consumption and low cost.

The cryptographic performance of the processors is measured by the built-in benchmark of the popular TrueCrypt utility, which uses AES-Twofish-Serpent "triple" encryption. It should be noted that this program is not only able to efficiently load any number of cores with work, but also supports a specialized set of AES instructions.

The atypical arrangement of processors in the diagram above is explained by the fact that Kabini and Richland, unlike all other processors participating in the testing, have support for the AES cryptographic instruction set. Accordingly, it helps them a lot in encryption tasks. And even the Sempron 3850, which in all tests before this was firmly in last place, was able to outperform the Celeron 1037U here.

To measure the speed of processors when compressing information, we use the WinRAR 5.0 archiver, with which we archive a folder with various files with a maximum compression ratio of 1.7 GB in total.

The big problem with the Socket AM1 platform lies in the fact that Kabini processors are equipped with only a single-channel DDR3 SDRAM controller. Therefore, in WinRAR, which, among other things, requires a high speed of the memory subsystem, representatives of the Kabini family do not look very good. For example, Athlon 5350 loses out to Celeron 1037U by almost 20 percent. However, at the same time, the older Socket AM1 processor manages to outperform Celeron J1900, whose memory controller, by the way, has two channels.

Gaming performance

The situation with the computing performance of Kabini desktop processors is generally clear. They can provide sufficient (by the standards of budgetary and energy-efficient solutions) speed of work in well-parallelizable simple counting algorithms. But some applications typical for home and office PCs of the entry level require different qualities from the CPU, therefore, when solving ordinary tasks, the Socket AM1 platform is not the best choice among the available options.

However, AMD processors usually have another trump card in their asset - the graphics core. Kabini has moved it to the latest GCN architecture, and if it proves to be able to provide acceptable gaming performance, Socket AM1 could be very interesting. However, in Kaveri, where integrated graphics have received decent performance, the GPU is based on six or eight compute clusters. In Kabini, there are only two such clusters, so there is no reason to expect that the Athlon 5350 and Sempron 3850 will be able to "pull" games in FullHD resolution at least with the minimum quality.

To estimate the relative performance of the graphics core of the heterogeneous Kaveri processor, we used the Futuremark 3DMark synthetic benchmark. Two subtests were used from the package: Cloud Gate, designed to determine the DirectX 10 performance of typical home computers, and the more resource-intensive Fire Strike, aimed at DirectX 11 gaming systems.

So, the Kabini graphics, belonging to the Radeon R3 class, turn out to be better than the GPUs built into the Bay Trail processors or the energy-efficient Celeron of the Ivy Bridge generation. However, it is inferior to the Haswell processor's GT1 graphics core, which is architecturally based on ten execution units, and is noticeably inferior to the Radeon HD 8470D from the A6-6400K processor.

However, 3DMark is a purely synthetic test, and it would not be entirely correct to draw any general conclusions based only on its performance. Therefore, let's see how the Kabini graphics core shows itself in real games. Considering the low potential of this core, the tests were run at 1280x720 with a choice of low image quality.

Already from these three examples, it is easy to understand that the integrated graphics of Kabini are not suitable for serious gaming use at all. In low resolution and at the lowest quality level, we get a terrible picture, but the fps level is barely close to what can be called acceptable. In other words, the lot of the Socket AM1 platform for entertainment use can be either undemanding casual or browser games, in which Kabini can really provide better graphics performance than inexpensive energy-efficient Intel processors.

We can end the conversation about the GPU built into Kabini. In the next generation of its energy efficient processors, Beema, AMD plans to roughly double the level of graphics performance. We will wait for the company to offer such processors for the desktop market, I want to believe that with them the creation of budget entry-level gaming systems will still become possible.

Video playback

The graphics core of Kabini processors can be used not only for 3D, but also to speed up video encoding and decoding. To do this, it inherited the VCE (Video Codec Engine) and UVD (Universal Video Decoder) functional blocks from full-fledged video cards. True, the VCE coding unit is currently only of theoretical interest; there are no popular and functional video transcoding utilities that would use its capabilities. But the UVD block is actively used by software players when decoding all common formats.

In order to test its effectiveness, we decided to look at the playback quality and the processor load when playing various H.264 video variants. The tests were carried out using the software player Media Player Classic - Home Cinema version 1.7.5 with the installed K-Lite Codec Pack 10.4.5 and with activated video decoding via LAV Filters 0.61.2.

The following graph shows the average load of the computational and graphic processor cores when playing a regular AVC FullHD video with a resolution of 1920 × 1080 and a frame rate of 25 fps. The bitrate of the test video is about 13 Mbps.

All test processors cope with the playback of ordinary FullHD video without any problems. This is not surprising. CPU and GPU load in any system remains at a low level. Consequently, even very inexpensive desktop processors have a good margin of power and can play more intricate video files without problems.

Let's complicate the task. The second test measured the load when playing AVC FullHD video with a resolution of 1920 × 1080 and a frame rate of 60 fps. The video bitrate is about 20 Mbps.

No critical problems arise here either, although the load on the graphics cores increases significantly. And while the Kabini processors have GPU utilization rates of up to 90 percent, they handle playback just fine. We did not observe any frame drops during testing.

Let's now see how the tested processors will cope with the playback of a video file encoded with the Hi10P profile using 10-bit color depth. The test video file has a resolution of 1920 × 1080, a frame rate of 24 fps and a bit rate of about 12 Mbps.

Support for hardware decoding of Hi10P video in modern GPUs has not yet been fully implemented. Therefore, most of the reproduction work falls on the computational processor resources. Which, however, cope with decoding without causing any complaints: their power is quite enough. Even the slowest processor in today's testing, the Sempron 3850, only slightly exceeds 50 percent.

And the final test is the playback of the increasingly popular 4K video. The resolution of the test video fragment is 3840x2160, the frame rate is 30 fps, the bit rate is about 100 Mbps.

This is where many low-cost processors have serious problems. Including Kabini. The Socket AM1 system shows a complete failure when playing 4K video: the processor load reaches 100 percent, and the user sees jerks and frame drops. For the sake of fairness, it should be noted that a similar picture is observed with the Bay Trail, this processor is also not suitable for playing videos with ultra-high resolution. But the Celeron processors belonging to the Ivy Bridge and Haswell generations show themselves quite differently: their built-in GPUs are capable of hardware decoding of 4K content, so viewing such video on systems based on them does not cause any difficulties. In summary, the Socket AM1 platform can be considered a suitable base for media players and HTPCs with some limitations.

Energy consumption

As tests have shown, from a performance standpoint, Kabini processors behave somewhat inconsistently. It is impossible to say that they are superior to Intel's energy efficient solutions. Yes, in some tasks their performance is higher, and such tasks are well-parallelizable algorithms for final rendering or video transcoding. But there are also opposite situations: under a typical office or home load, Socket AM1 processors are outperformed by both Celeron J1900 and Celeron 1037U.

However, it should be borne in mind that good energy efficiency is usually expected from processors of this class. And here Kabini can show themselves on the positive side. The underlying Jaguar microarchitecture is initially focused on low consumption, and processors based on it are used even in tablets. All this gives hope that the Socket AM1 platform will be able to fully compete with competing offers in terms of its efficiency. Let's check.

The following graphs, unless otherwise specified, show the total consumption of systems (without a monitor), measured at the outlet from the outlet to which the test system's power supply is connected, and representing the sum of the energy consumption of all components involved in it. The total indicator automatically includes the efficiency of the power supply itself, however, given that the power supply model we use, Corsair AX760i, is certified 80 Plus Platinum, its effect should be minimal. During the measurements, the load on the computational cores of the processors was created by the 64-bit version of the LinX 0.6.4 utility. The Furmark 1.13.0 utility was used to create a load on the graphics cores. For a correct assessment of power consumption in various modes, we use all available energy-saving technologies: C1E, C6, Enhanced Intel SpeedStep and Cool "n" Quiet.

In terms of idle consumption, platforms built on systems-on-a-chip are in the lead. They are distinguished by a single-chip design that does not require additional hubs - system logic sets, which allows them to provide high energy efficiency at rest. This means that from the point of view of economy, Socket AM1 systems can really turn out to be a good option. In the idle state where real systems spend most of the time, the Athlon 5350 and Sempron 3850 outperform even the Bay Trail-D.

However, under the computational load, the consumption picture of desktop Kabini no longer looks as favorable. Athlon 5350 turns out to be a noticeably more power hungry processor than Celeron 1037U and Celeron J1900. In terms of its consumption under load, it loses only to full-fledged desktop models, the performance of which is several times higher.

On the other hand, the GPU built into Kabini is quite economical. The only pity is that its performance is not enough for gaming use - it could have turned out to be a very interesting option.

It is curious that with the simultaneous load on both computing and graphics capacities Athlon 5350 compares with Celeron 1037U in consumption. This is because Intel HD Graphics is significantly less power efficient than the GCN graphics used in Kabini. However, in terms of total power consumption under load, the Bay Trail-D - Celeron J1900 wins by a large margin. This economical Intel processor allows you to build a desktop system that consumes no more than 35 watts in any situation. Even the youngest quad-core Kabini, Sempron 3850, consumes 10 W more under the same conditions.

conclusions

Summing up, we can make an unambiguous conclusion that the new Socket AM1 Kabini are the best AMD processors in terms of combination of consumer characteristics today. However, they occupy this position in the range of the company's products not so much due to some of their indisputable advantages, but because AMD simply does not have any other balanced and attractive offers for the broad masses of users. Kabini, given their positioning, have clear advantages.

The Socket AM1 platform is aimed by the manufacturer to occupy the initial market segment due to a good combination of performance and price, as well as performance and power consumption. Now in this segment, small-format motherboards equipped with integrated Intel Bay Trail processors or energy-efficient Intel Celeron are fixed. AMD, on the other hand, wants to squeeze Intel's options with its new platform, offering better characteristics and the possibility of a subsequent upgrade. Although the arguments put forward by AMD at times seem controversial, the overall potential of Kabini in the desktop market is hard to question.

In announcing the desktop Kabini, AMD put forward the slogan "four cores for a penny", and it is surprisingly aptly reflects the essence of these CPUs. Combining four cores with the Kabini microarchitecture, Socket AM1 processors can demonstrate relatively good performance in multi-threaded environments. In these situations, such processors really outperform their direct competitors in speed: the quad-core Bay Trail-D and the dual-core energy-efficient Ivy Bridge. Of course, under a typical workload for inexpensive desktop systems, the performance of Kabini is far from the best in its class, but in fact, the responsiveness of such processors in office and Internet applications is quite sufficient, and more is not needed for many users.

The situation is not bad with energy consumption. On the one hand, under high load, the energy efficiency of Intel's Bay trail-D is better, but on the other hand, the Kabini system-on-a-chip can offer very low power consumption in idle and when running graphics, which may well convert into good average efficiency. In general, the Socket AM1 platform can certainly be placed in tight cases and equipped with low-power power supplies. Hopefully, Kabini-compatible passive cooling systems will soon appear on the market as well.

Another advantage of Kabini could well be the integrated graphics core, it is really clearly better in these processors than in the main competitors. But, unfortunately, it is still too weak to provide at least the minimum level of performance in modern games. The media engine does not look outstanding either: it turned out to be incompatible with the increasingly popular AVC video in 4K resolution.

Nevertheless, in the end it turns out that the Socket AM1 platform may be the best choice in a fairly large number of situations when it comes to building a budget system. This is exactly what AMD was counting on: first of all, Kabini is for those who like to save money. Of course, it's a shame that the four cores of Jaguar seriously fall short of the performance of the dual-core Haswells of the Celeron class, but this is unlikely to prevent the Kabini processors from fitting well in the lower part of the desktop segment. Their main advantage is that, at a minimal cost, they have no obvious disadvantages, which means that the Socket AM1 platform can become a universal solution for many users.

Socket compatibility for Socket AM2, AM2 +, AM3 and AM3 +

Socket AM3 +
Socket AM3 + is a continuation of Socket AM3, mechanically and electrically compatible with Socket AM3 (despite a slightly larger number of contacts - 942, it can also be called SocketAM3b in some sources). Designed to support new AMD processors based on Zambezi core with Bulldozer architecture (for example, AMD FX 8150). Socket AM3 + is compatible with Socket AM3 processors and coolers for Socket AM2 / AM3.

Socket AM3
Socket AM3 is a further development of Socket AM2 +, its main difference lies in the support of DDR3 memory by motherboards and processors with this type of socket. Socket AM3 processors have a memory controller that supports both DDR2 and DDR3, so they can work in Socket AM2 + motherboards (processor compatibility should be checked on the CPU Support List on the motherboard manufacturer's website), but the opposite situation is impossible, Socket AM2 and Socket AM2 + processors in Socket AM3 boards do not work.

Socket AM3 motherboards support DDR3 memory frequency from 800 to 1333 MHz (including with ECC). With currently manufactured Socket AM3 processors, PC10600 memory will operate at the nominal frequency of 1333 MHz only if one module is installed per channel, and when two modules are installed on each channel of the memory controller (when there are three or four memory modules in total) their frequency forcibly reduced to 1066 MHz. Registered memory is not supported, ECC memory (without Registered) is supported only by Phenom II processors for this socket. The memory architecture is dual-channel, therefore, to achieve optimal performance, it is necessary to install two or four (preferably identical in pairs) memory modules in accordance with the instructions for the motherboard.

Socket AM2 +
Socket AM2 + is an upgraded version of Socket AM2. The differences include support for HyperTransport 3.0 technology up to 2.6 GHz and improved power circuits.
Basically, all Socket AM2 processors work fine in all Socket AM2 + motherboards (there are exceptions related to individual technical features of some motherboards). Not all Socket AM2 motherboards support Socket AM2 + processors (compatibility in each specific case should be found out on the motherboard manufacturer's website), secondly, a decrease in the HyperTransport frequency leads to a noticeable drop in processor performance compared to Socket AM2 + motherboards. Also, when using Phenom Soket AM2 + processors, the boards allow using DDR2 RAM (for example, PC-8500) at the rated frequency without overclocking (when installed one module per channel).

Hello readers of my blog about hardware. In this article, I wanted to consider which processors are suitable for the am3 and am3 + socket. Despite the fact that this connector from AMD came out more than 7 years ago, it is still in demand in the market, since the release of new chips for AM4 has greatly reduced the prices for the FX-8xxx, which, moreover, are also great at chasing.

If you want to know which processors are suitable for 1151 -, but here we will consider the supported products that can be installed in socket am3 +. We will also briefly touch upon some characteristics of the chips, such as the highest performing FX-9590 and the popular FX-8300.

List of supported chips

If you look at the official statistics, then AM3 + in theory has no compatibility with AM3, however, the outdated chips work fine on a newer socket, without any hardware limitations in terms of overclocking. The table will include both new and old CPU models, among which you are likely to find the best processor for gaming.

Vishera (32 nm):

Bulldozer (32 nm):
As you can see, the FX generation has 2 incarnations, which include 2 architectures, and Vishera is a modified and improved version of Bulldozer. Both options will run smoothly on any motherboard.

Models for AM3 are also suitable for AM3 +.

Their lineup looks like this:
Which processor for your PC? Try to look for the most "fresh" solution, i.e. AMD FX. Let's say right away that the 4-core FX-4100 is not the best choice for creating a system, since there are more advanced FX-8xxx, in particular the 8300, which can easily run up to 4.8 GHz on a 970 chipset using cooling, the level of Zalman CNPS10 Optima or Deepcool Gammaxx 300.

New chips can still be found on sale at very attractive prices and we recommend purchasing the OEM versions, since they are cheaper than BOX and are not inferior in performance. you can read about the main differences between BOX and OEM.

A few words about AM4

In 2016, a completely new processor socket for AMD Ryzen processors - AM4 was introduced to the market. Unlike the previous versions (AM3 +, AM3, AM2 +, AM2) this socket is completely new and does not have backward compatibility with outdated processors from the Red ones. However, it supports much more interesting chips that are relevant at the time of 2018:
The list includes models based on the Zen and Zen + architecture, which are backward compatible with each other and work great on motherboards with A320, B350, B450, X370 and X470 chipsets.

Optimal models

If you need the most balanced processor of the AM3 + family, then we recommend taking a closer look at the already mentioned FX-8320 with a base frequency of 3.5 GHz, which can be raised to 4 in Turbo Boost without any problems, or manually increased to 4.5 if there is good cooling.

Do you have a board based on the top-end 990FX chipset? Try the FX-9590, which runs at 4.7 GHz in stock and can squeeze out up to 5 GHz, but on one condition - a 220 W heat pack. And this speaks of the very "hot" nature of the chip.

And now for the modern AM4 models. The best solution for an entry-level multimedia system would be Ryzen 5 2400G with integrated Vega 11 video core, whose performance is comparable to the GeForce 1030 GT
As a universal solution, we want to offer Ryzen 5 1600, which has 6 cores and 12 threads, as well as a low thermal package of 65 W and an excellent performance margin. This stone will provide maximum immersion in any game or program.

Introduction The coming summer promises to be a truly hot season. And if from a meteorological point of view, this forecast may not be justified due to the action of powerful cyclones, then on the processor market everything has already been determined absolutely precisely. Both top players, AMD and Intel, have chosen summer to upgrade their high-performance platforms. So, Intel in the middle of summer will bring to the market processors with a fundamentally new Core microarchitecture, and AMD will focus on promoting the Socket AM2 platform, which provides support for DDR2 SDRAM, throughout the summer season.
Although the most anticipated processors at the present time should be considered the CPUs of the Intel Core 2 Duo family, also known by their codename Conroe, AMD, according to the tradition that has developed over the past few years, has outstripped the competitor and will begin mass deliveries of its updated processors for the Socket AM2 platform on June 1 ... That is why today we will take a closer look at new products from AMD, postponing the publication of Core 2 Duo reviews for some time, until their official announcement.
Despite the impending release of very promising Intel processors, AMD's Socket AM2 platform is attracting a lot of attention. AMD delayed the transition to DDR2 SDRAM to the last, because the K8 processor microarchitecture, which includes an integrated memory controller, benefits primarily not from the memory bandwidth, but from its low latency, which the DDR2 SDRAM on the market cannot boast of. Nevertheless, today the speeds of DDR2 memory have increased so much that transferring the Athlon 64 processors to work with this type of memory can theoretically yield tangible dividends in the form of performance gain. Although the first tests of engineering samples of the new platform from AMD did not reveal any particular advantages of it, now we are talking about serial processors and motherboards. This is the main intrigue of this material. Many fans of AMD processors would like to believe that Socket AM2 processors will be able to compete on equal terms with Intel Core 2 Duo.
In addition, the updated AMD processors receive a new revision core, which, in addition to supporting new types of memory, has some cosmetic alterations, which at the same time also increase the attractiveness of the Athlon 64 family of processors. AMD solutions to the "enemy camp". But it is too early to make hasty conclusions, especially since some improvements in the K8 processors may be in great demand in some cases. So, let's take a closer look at AMD's Socket AM2 processors and try to predict how attractive they will be to potential consumers.

Revision F core: the basics

For use in new processors designed for the Socket AM2 platform, AMD has developed an updated core with the K8 microarchitecture, which received the revision number F. Thus, all dual-core and single-core AMD processors with an integrated memory controller supporting DDR2 SDRAM will be based exclusively on this core. ...
The main innovation in the microarchitecture introduced by the core of the new revision is support for DDR2 memory. In the new core, AMD simply replaced the memory controller, since the Athlon 64 microarchitecture allows making such changes without any problems. At the same time, the new memory controller for Athlon 64 processors is not backward compatible with DDR SDRAM. This means that from today, DDR memory can be classified as an obsolete solution. Today's platforms from leading processor manufacturers AMD and Intel are now unanimous and require DDR2 SDRAM. Obviously, this should affect the cost reduction of such memory, and in the very near future the cost of DDR2 SDRAM will be set at a lower level than the price of DDR memory modules of the same size.
Returning to the issue of DDR2 SDRAM support by the revision F core memory controller, it should be noted that it officially supports memory up to 800 MHz. In other words, AMD managed to implement DDR2-800 SDRAM support in its platforms earlier than Intel. Naturally, new AMD processors are also compatible with slower DDR2 memory with frequencies of 667 or 533 MHz. But, taking into account the fact that low latency of memory is first of all important for the K8 architecture, it is the use of DDR2-800 SDRAM that can give the maximum effect in terms of performance.
It should be noted that traditionally the memory controller of the new core is equipped with a slightly larger number of dividers for the DDR2 operating frequency than is indicated in the official specification. Due to this, some motherboards will be able to provide operation of Athlon 64 processors for Socket AM2 systems even with DDR2-1067 SDRAM, without overclocking the clock generator. But so far AMD does not declare work with memory faster than DDR2-800 in its official documents.
In addition to support for DDR2 SDRAM, the revision F core can boast of some additional innovations. For example, Athlon 64 processors for the Socket AM2 platform now support the virtualization technology known under the codename Pacifica. This is a symmetrical response to Intel VT technology introduced in Intel Presler processors.
An equally important circumstance connected with the transfer of AMD processors to the revision F core was the decrease in their power consumption. Despite the fact that for the production of processors AMD continues to use the old technological process with production standards of 90 nm (with SOI and DSL technologies), Socket AM2 processors have lower heat dissipation and power consumption than their Socket 939 counterparts. Formally, the transfer of dual-core Athlon 64 X2 processors to a new core allowed us to lower the maximum heat dissipation limit by 19%, from 110 to 89 W, and the maximum heat dissipation of single-core Athlon 64 processors was reduced by 30% thanks to the revision F core - from 89 to 62 W.
The mentioned increase in efficiency is an equally important improvement of the new core, along with the transition to support for DDR2 memory. Especially in light of the fact that the ratio "performance per watt" is currently actively promoted by CPU manufacturers as the main metric for assessing the consumer qualities of their products.
However, the indicated decrease in heat dissipation of mainstream AMD processors is not all. The fact is that with the release of the Socket AM2 platform and the transition of the manufacturer to the use of revision F cores on the basis of their CPUs, it became possible to release additional energy-efficient (Energy Efficient) processor lines. AMD is going to offer consumers two options for energy-efficient CPUs: with a maximum heat dissipation, limited to 65 and 35 watts. Obviously, processors with a maximum heat dissipation of 65W will compete with Conroe in terms of thermal and electrical characteristics, and 35W units will be intended for use in small, quiet and economical systems. AMD does not plan to use any special manufacturing technologies for the production of energy-efficient processors. Such CPUs will be mined by simple selection of crystals among all processors of revision F.
The transfer of AMD processors to the Socket AM2 platform will be massive. For the new platform, both dual-core Athlon 64 X2 processors, single-core Athlon 64 and budget Sempron processors will appear simultaneously. Therefore, the kernels of revision F will simultaneously exist in several guises. Possible options and their formal characteristics are shown in the table below.


And this is how the Athlon 64 X2 processor core, revision F.

It should be noted that, despite the appearance of support for DDR2 SDRAM, the core of revision F does not contain any cardinal improvements in terms of microarchitecture. Since the release of the first Athlon 64 processor family, AMD has avoided making any changes directly to the decoders or core executors. That is, roughly speaking, so far we are observing the development of the K8 architecture only along the extensive path of making small improvements. And that was enough for Intel to compete successfully. But now the situation is changing. Coming out this summer, the Intel Core 2 Duo processors have a fundamentally new microarchitecture, which is distinguished by the ability to execute up to 4 instructions per cycle. And it will be quite difficult for AMD processors to compete with them, given that they do not have the same theoretical peak performance. From this position, the core of revision F, despite all the innovations present in it, is somewhat disappointing. To be honest, we would like more from him, first of all improvements at the microarchitecture level. But AMD engineers have nothing to offer us yet.

Socket AM2 platform

Let's take a closer look at what the new Socket AM2 platform offers to the user, in addition to DDR2 SDRAM support.
First of all, it should be noted that formally Socket AM2 is a 940-pin processor socket. At the same time, Socket AM2 processors are neither logically nor electrically compatible with the old Socket 939 and Socket 940 sockets. To protect users from incorrect installation, Socket AM2 processors cannot physically be installed in old motherboards, they are located differently legs.

A positive moment in the transition to Socket AM2 is that from now on AMD will offer a single platform for expensive dual-core and single-core budget processors. The same Socket AM2 motherboards can work with both Athlon 64 X2 and Athlon 64 and Sempron processors.
However, the introduction of a new processor socket does not yet sign the death warrant for the old ones. AMD promises to continue to support and ship Socket 939 products as long as there is consumer interest in the platform.
Socket AM2 also sets new requirements for motherboards in terms of maximum power consumption and heat dissipation of processors. Although we talked about the fact that new CPUs with a revision F core can boast of lower power consumption, the platform's capabilities to support electrically powerful processors have been increased. Now the upper limit on the consumed current is set at 95 A versus 80 A provided by Socket 939 motherboards. All this can make it possible to use processors that consume up to 125 W, while the maximum power consumption of the Socket 939 CPU was limited to 110 W.
Together with a new, more powerful power supply scheme for Socket AM2 processors, motherboards offer a new mechanism for mounting the cooler. Now the frame on which the cooler is fixed is screwed to the motherboard with not two, but four bolts. But the fixing "teeth" on the frame remained in their old places.

This means that Socket AM2 motherboards can allow the use of old cooling systems, provided that they were mounted on a regular frame. The same heat removal systems that were screwed directly to Socket 939 motherboards will not be able to be used on new platforms without modification.

Processors for Socket AM2

In the table below, we present the full list of Socket AM2 processors that will be available for sale after June 1st.

It should be noted that the correspondence between frequency, cache memory size and CPU rating for the Socket AM2 platform is the same as for Socket 939 processors. On the one hand, this will allow users to more easily navigate the characteristics of the new processors, but on the other hand, it makes it clear that AMD does not expect a noticeable increase in performance from the transition to the new platform and processor core.
I would like to draw your attention to the fact that support for the fastest memory, DDR2-800 SDRAM AMD is declared only for dual-core processors. Single-core CPUs, according to the official specification, can only work with DDR2-667 memory. This is quite logical, given the increased demands of dual-core CPUs for memory bandwidth, at least due to the fact that RAM is directly involved in solving issues of core cache coherence.
The line of Socket AM2 processors has been significantly expanded due to the appearance of energy efficient processors with two new thermal packages - 65 and 35 W. These processors do not have such high frequencies as their "full-fledged" counterparts and are somewhat more expensive. However, they can be very attractive options for a variety of applications, including small, quiet computers. However, the preferences of the bulk of consumers, including enthusiasts, are unlikely to be on the side of these processors. In other words, we do not yet expect widespread adoption of energy efficient CPUs.
However, it should be remembered that processors with reduced thermal efficiency can be easily identified by their markings. While the third letter in the line of marking conventional processors is "A", for CPUs with a thermal package of 65 W it will be changed to "O", and the most economical processors with a heat dissipation limited to 35 W will be marked with the letter "D" ...
Unfortunately, the appearance of Socket AM2 processors will not contribute much to the rise in popularity of dual-core processors from AMD. The transition to a new platform, although it expands the range of dual-core offerings of the company, does not entail a decrease in prices for processors with two cores. All Athlon 64 X2 processors will continue to sell for over $ 300, which is unlikely to have a positive impact on their popularity. Especially considering the fact that Intel, in the light of the imminent appearance of a CPU with a new microarchitecture Core, has thrown a large number of cheap dual-core processors onto the market. For example, the cost of Intel's junior dual-core processor has already dropped well below the $ 150 mark. So from these positions, it is Intel that should be considered the main locomotive promoting dual-core CPUs to the market.

Test processors: Athlon 64 FX-62 and Athlon 64 X2 5000+

To test the performance of the new Socket AM2 platform, AMD sent us two processors: Athlon 64 FX-62 and Athlon 64 X2 5000+. The first of them is a dual-core processor aimed at gamers who are willing to do anything (financially) to achieve maximum performance, the second is the senior dual-core processor in the Athlon 64 X2 line.
Athlon 64 FX-62 has the highest frequency among new and old AMD CPUs at 2.8 GHz. Moreover, it even caught up with the frequency of the single-core Athlon 64 FX-57! However, this did not pass without a trace for him: the maximum heat dissipation of the novelty is 125 W, which can be called a kind of record. There are no other equally hot processors among AMD products yet.

Diagnostic utility CPU-Z gives the following information about Athlon 64 FX-62.

It should be noted that the nominal voltage of the Athlon 64 FX-62 is 1.35-1.4 V, which is more than that of other dual-core CPUs in the Athlon 64 X2 line.
All this clearly indicates that the frequency potential of 90 nm cores with the K8 microarchitecture is coming to an end. However, the results of the Athlon 64 FX-62 overclocking indicate that if we close our eyes to the growing power consumption, we can achieve more.
Thus, our test processor, when its supply voltage was increased to 1.5 V, was able to operate stably at 3075 MHz, obtained as 15 x 205 MHz (Athlon 64 FX processors have a variable multiplier).

At the same time, heat was removed from the processor using a completely ordinary air cooler from AVC (article Z7U7414002).

It must be said that overclocking a dual-core Athlon 64 FX-62 to a frequency higher than 3.0 GHz without using special means for cooling is quite an impressive fact. Typically, all FX series processors were air-cooled and only allowed to increase their frequency by about 200 MHz. So, if desired, AMD will be able to increase the nominal frequencies of its dual-core processors up to 3 GHz. The only thing that can prevent this venture from being carried out is the excessively increasing power consumption and heat dissipation of the CPU. For example, the power consumption of our test copy Athlon 64 FX-62, overclocked to 3.075 GHz and working under full load, according to the measurement results, amounted to 192 W (!), Which clearly does not fit into the requirements that AMD itself set for the Socket platform AM2.
The second processor from our laboratory, Athlon 64 X2 5000+, has a nominal clock frequency of 2.6 GHz, but is inferior to the FX-62 in terms of L2 cache size. The cache memory of each of its cores is 512 KB.

The CPU-Z utility detects this processor as follows.

It should be noted that all dual-core processors of the Athlon 64 X2 line, including the 5000+ model, have a supply voltage reduced to the range of 1.3-1.35 V. This, in particular, allows such processors to fit into the thermal package limited by the maximum heat dissipation in 89 Watt
Comparison of the electrical characteristics of new Socket AM2 processors measured in practice allows to get a very interesting picture. As always in our tests, the processor load when measuring the maximum power consumption level was performed by a specialized utility S&M, which can be downloaded here. As for the measurement technique, it, as usual, consisted in determining the current passing through the processor power circuit. That is, the figures below do not take into account the efficiency of the CPU power converter installed on the motherboard.

We are already so used to the fact that one of the characteristics of processors with the NetBurst microarchitecture is high heat dissipation. So the figures shown in the diagram are capable of plunging into a slight shock. But you can't argue against facts. AMD's senior processor, Athlon 64 FX-62, has a slightly higher power consumption and heat dissipation than Intel's senior dual-core processor, Pentium Extreme Edition 965, which is based on the Presler C1 revision core. The older processors in the mainstream dual-core lines, Athlon 64 X2 5000+ and Pentium D 960, are now showing approximately the same heat dissipation level. Thus, older AMD processors can no longer be awarded the title of more economical. The latest CPUs from Intel, based on the latest Presler core revision, are clearly no worse in this parameter. Thus, the Socket AM2 platform has acquired increased current and heat dissipation tolerances for processors for a reason.
However, let's get back to the Athlon 64 X2 5000+ processor, namely, let's talk about its overclocking potential. Overclocking this CPU has to be done by increasing the frequency of the clock generator, its multiplier is fixed at the top. However, this does not interfere with achieving high results. By increasing the supply voltage of our test unit to 1.5 V, we were able to achieve stable operation at 2.99 GHz.

The results of overclocking two Socket AM2 processors using the simplest air cooler allow us to say that the frequency potential of CPUs with revision F core has become somewhat higher than that of previous AMD processors. Thus, the Socket AM2 platform can be quite interesting for overclockers.

Chipsets

Since the communication of the logic sets and all processors with the K8 microarchitecture is carried out using the HyperTransport bus, and the memory controller is integrated into the CPU, the transition of the Athlon 64 family to use a new socket and DDR2 SDRAM memory does not require the use of any special logic sets. All those chipsets that were used in Socket 939 motherboards can be successfully used in Socket AM2 motherboards.
However, despite this, NVIDIA company, which at the moment can be considered the leading supplier of chipsets for AMD processors, marked the release of a new platform from AMD with the announcement of new sets of system logic for it. New chipsets of the NVIDIA nForce family (nForce 590, nForce 570, nForce 550) are positioned by the manufacturer as "specially designed for new AMD processors." However, there is nothing special from the point of view of support for processors in these chipsets, they are remarkable only for their advanced capabilities. The simultaneous announcement of new NVIDIA logic sets and the Socket AM2 platform is just a marketing step.
However, the transition to the new AMD platform will still require changing the motherboard. In this regard, the new chipsets are quite popular, because most users will certainly want to get a new board with more features. It is for this category of consumers that the new chipsets from NVIDIA are designed.
The new NVIDIA nForce family of chipsets includes four products aimed at a disjointed target audience.

All these chipsets are built on the same element base, which is based on the nForce 570 chipset. It should be considered the starting point from which the rest of the products - nForce 590 and nForce 550 - stand.
The NVIDIA nForce 570 SLI chipset is a single-chip solution that can be called a further development of the nForce 4 SLI.

This chipset supports SLI mode, but only on the PCI Express x8 + PCI Express x8 scheme.
A similar chipset NVIDIA nForce 570 Ultra is the same product, but without the option to activate the SLI mode.

For the most "advanced" part of the gaming community, NVIDIA has prepared the nForce 590 SLI chipset, which is capable of supporting SLI modes according to the PCI Express x16 + PCI Express x16 scheme. In this implementation, to support the second PCI Express x16 graphics slot, the chipset includes an additional microcircuit connected to the processor and MCP via the HyperTransport bus with a width of 16 bits in each direction and a frequency of 1 GHz.

As for the budget NVIDIA nForce 550 chipset, this is the same nForce 570 Ultra, but with somewhat reduced capabilities.

The formal characteristics of the new nForce chipsets are summarized in the table below:

A study of the characteristics of the new NVIDIA chipsets for the Socket AM2 platform shows that they do not differ much from the previous generation of nForce4 chipsets. In fact, there are only three major improvements in the new chipsets:

Dual port Gigabit Ethernet controller;
Increase in the number of SATA channels to six;
The long-awaited appearance of High Definition Audio.

I must say that despite such a small list of improvements, NVIDIA gives the new chipsets a huge step forward, which is facilitated by both the marketing protrusion of some chipset features and the additional features being developed implemented at the software level.
Without going into details, let us note the main technologies present in the chipsets, which are the subject of special pride for NVIDIA engineers:

LinkBoost... Automatic overclocking of PCI Express x16 buses to increase the throughput between the GeForce video cards installed in the system;
SLI-Ready Memory... Another name for the previously announced Enhanced Performance Profile technology, which allows the use of memory modules with extended SPD content, in which, in addition to the main timings, the optimal voltage of the modules and the values \u200b\u200bof the secondary parameters are preserved.
FirstPacket... A technology that allows you to assign high priority to network packets generated by specific applications. NVIDIA uses it to reduce pings in gaming applications.
DualNet... Dual-port network controller for chipsets allows both ports to be used separately or together for one connection.
TCP / IP Acceleration... Part of the TCP / IP packet processing routine traditionally performed by the network card driver is transferred to the hardware capabilities of the logic set.
MediaShield... The six-port Serial ATA II controller of the chipset allows the formation of one or several RAID arrays of levels 0, 1, 0 + 1 and 5.

In addition, along with motherboards based on the new nForce 590/570/550 chipsets, NVIDIA plans to supply a new utility called nTune 5.0, which has now acquired new opportunities for monitoring and fine-tuning the system.
One of the first motherboards based on the NVIDIA nForce 590 SLI chipset was the ASUS M2N32-SLI Deluxe, which we used in our tests.

How we tested

To test the performance of the new Socket AM2 AMD processors, we used the following set of hardware:

Processors:

AMD Athlon 64 FX-62 (Socket AM2, 2.8GHz, 2x1MB L2);
AMD Athlon 64 FX-60 (Socket 939, 2.6GHz, 2x1MB L2);
AMD Athlon 64 X2 5000+ (Socket AM2, 2.6GHz, 2x512KB L2);
AMD Athlon 64 X2 4800+ (Socket 939, 2.4GHz, 2x1MB L2);
Intel Pentium Extreme Edition 965 (LGA775, 3.76GHz, 2x2MB L2).
Intel Pentium D 960 (LGA775, 3.6GHz, 2x2MB L2).

Motherboards:

ASUS P5WD2-E Premium (LGA775, Intel 975X Express);
ASUS M2N32-SLI Deluxe (Socket AM2, NVIDIA nForce 590 SLI);
DFI LANParty UT CFX3200-DR (Socket 939, ATI CrossFire CFX3200).

Memory:

2048MB DDR400 SDRAM (Corsair CMX1024-3500LLPRO, 2 x 1024 MB, 2-3-2-10);
2048MB DDR2-800 SDRAM (Mushkin XP2-6400PRO, 2 x 1024 MB, 4-4-4-12).

Graphics card: PowerColor X1900 XTX 512MB (PCI-E x16).
Disk subsystem: Maxtor MaXLine III 250GB (SATA150).
Operating system: Microsoft Windows XP SP2 with DirectX 9.0c.

Testing was carried out with motherboards BIOS Setup set to maximum performance.

DDR2 versus DDR: was there any point

Anticipating the performance tests of new AMD processors for the Socket AM2 platform, we decided to pay special attention to finding out what the transfer to DDR2 SDRAM can give in terms of performance for Athlon 64 processors. After all, it is no secret to anyone that platforms based on AMD CPUs are very critical of the memory subsystem latency. And the transition from DDR to DDR2 SDRAM, although it promises a significant increase in throughput, does not give a gain in latency.
To get practical data that would allow us to draw some conclusions about the benefit that AMD received from using DDR2 SDRAM in their systems, we collected two similar systems with DDR and DDR2 memory and compared their performance at different timings and different memory bus frequencies. Athlon 64 FX-60 for Socket 939 and Athlon 64 FX-62 slowed down to 2.6 GHz for Socket AM2 were used as central processors during the tests. Note that for these tests we used 512 MB memory modules, that is, the total amount of memory in the test systems was 1 GB.
First of all, let's take a look at the synthetic benchmarks that measure practical memory bandwidth and latency.

The results obtained in practice confirm the theoretical speculations. DDR2 SDRAM has a higher bandwidth than ordinary DDR memory, which is higher the higher its frequency. But in terms of latency, the picture is completely different. Only DDR2-800 SDRAM with rather aggressive (for such a frequency) 4-4-4 timings can compete with DDR400 SDRAM operating at minimum delays of 2-2-2. DDR2-667 SDRAM with the lowest possible timings of 3-3-3 can achieve only about the same practical latency as DDR400 with 2.5-3-3 delays, it cannot compete with fast DDR SDRAM. As for DDR2-533 SDRAM, from the point of view of latency this memory is guaranteed to be worse than any DDR400 SDRAM.

The SiSoftware Sandra 2007 results are quite consistent with the data we obtained using another test, Sciencemark 2.0. In fact, we can already say that only those owners of Socket AM2 platforms who will use either DDR2-800 SDRAM or fast DDR2-667 memory with 3-3-3 delays in their systems can get a performance gain. The performance gain in all other cases remains questionable and will depend primarily on the nature of the tasks being solved.
From testing the parameters of the memory subsystem, let's move on to considering the speed of work in complex tests.

The SuperPi test only exacerbates the above statements. Indeed, the Socket AM2 platform demonstrates higher performance than a Socket 939 system with DDR400 memory with 2-2-2 delays only if it uses DDR2-800 SDRAM.

Some tasks demonstrate a rather weak dependence on the speed of the memory subsystem. Nevertheless, the low efficiency of DDR2 SDRAM compared to fast DDR400 SDRAM can be noticed here as well.

The speed of the WinRAR archiver depends heavily on the performance of the memory subsystem. In this case, we see that this task is quite sensitive to the growth of throughput. But despite this, only DDR2-800 with 4-4-4 timings show a slightly higher result than the Socket 939 platform with 2-2-2 latencies demonstrates.

The same can be said when looking at the performance in games. Even the slowest DDR400 memory is better than some types of DDR2 SDRAM.
So, answering the question posed at the beginning of this section, it can be argued that there is no direct sense in increasing the platform performance in switching to DDR2 SDRAM. Another thing is that the transition to support for a newer memory standard can be useful from the point of view of future prospects. The development of DDR SDRAM has come to an end, both manufacturers and JEDEC have focused on developing fast memory standards based on DDR2. That is why AMD's choice should be considered correct. The company waited for the moment when DDR2-800 SDRAM became widely available on the market, which did not reduce platform performance, and switched to a new memory standard, looking into the future. By the way, a significant advantage of DDR2 memory compared to DDR SDRAM in light of the imminent release of the new generation Windows Vista operating system should be considered the better availability of large memory modules.

Performance

Synthetic tests: PCMark05, 3DMark06 and ScienceMark 2.0

First of all, we decided to check the performance of the processors in question using common synthetic tests.

It should be noted that there is nothing fundamentally new in the results obtained. As shown above, switching AMD processors to DDR2 SDRAM gives a small performance gain. Therefore, the high level of performance of the new CPU Athlon 64 FX-62 is primarily due to its high clock frequency of 2.8 GHz. The performance of the Athlon 64 X2 5000+ processor is in some cases inferior to that of the Athlon 64 FX-60, since, despite the same clock frequency, this CPU has half the cache memory size. However, in the tests for which the cache memory size is not important, Athlon 64 X2 5000+ can outperform any Socket 939 CPU, since in the tested configuration it is equipped with high-speed DDR2-800 memory.

Overall performance

We measured overall performance in digital content creation and office applications using the SYSMark 2004 SE benchmark, which also makes heavy use of multithreading.

When dealing with digital content, AMD processors significantly outperform competing Intel CPUs. As for the new Socket AM2 platform, it does not present any surprises to us in this case.

In office applications, the amount of cache memory is of great importance. Therefore, the Athlon 64 X2 4800+ processor for Socket AM2 systems outperforms the Athlon 64 X2 5000+. I would also like to note the rather high results shown in this benchmark by the Intel Pentium D 960. As you can see from the diagram, it is inferior in performance only to the AMD FX series processors, which are distinguished by a much higher price.

Audio and video encoding

When encoding audio and video using the DivX, iTunes and Windows Media Encoder codecs, we can observe a rather tangible advantage of the new Socket AM2 platform. Streaming video encoding is a challenge that responds well to increasing memory bandwidth. Accordingly, in these tasks the speed of Socket AM2 processors turns out to be higher than the speed of similar Socket 939 processors by about 2-4%.
Apple Quicktime is less enthusiastic about the new platform. In its operation, the Socket AM2 Athlon 64 4800+ processor even lags slightly behind its Socket 939 counterpart. However, in any case, there are no dramatic differences in performance, even when working with streaming data.

Image and video processing

Until recently, the Intel Pentium Extreme Edition processor remained the unrivaled leader in Adobe Photoshop and Adobe Premiere. But the release of the high-speed AMD Athlon 64 FX-62 processor changed this state of affairs. Now it is this processor from AMD that is named the fastest product for image processing and non-linear video editing.

Performance in 3ds max 7 and Maya

Unfortunately, the increase in the frequency to 2.8 GHz for the Athlon 64 FX-62 is not enough to compete with the Pentium Extreme Edition 965 in the final rendering in 3ds max. The thing is that rendering is a well-parallelized task that can fully load all four virtual cores that Intel's top processor has. However, when rendering in Maya, this picture does not repeat itself, the senior dual-core processors from AMD are leading in this package.
As for the effect of using DDR2 SDRAM by AMD processors, in this case we can speak of its absence or even negativeness. In any case, the final rendering is not a task for the sake of which AMD supporters should move to the new platform.

3D games

Theoretically, you can get a rather noticeable performance gain from switching to DDR2 memory in games. The fastest DDR2-800 SDRAM can provide a visible increase in speed, reaching 6-7% in some games. However, we are not talking about the qualitative superiority of the new platform. At the same time, preliminary test results of the promising Conroe processor show that it will provide a qualitative leap in performance for Intel processors in gaming applications. In other words, although AMD processors continue to maintain a confident leadership in games, in the near future this balance of forces can easily change. And supporters of the AMD platform need to be mentally prepared for such a turn of events.

Other applications

Since the performance of the Socket AM2 platform in comparison with the performance of desktop CPUs supporting DDR SDRAM seems to be a very interesting question to study, we decided to add several more common programs to the number of test applications.
Using the 7-zip archiver, which supports multithreading very efficiently, we measured the compression and decompression speed of the data.

We assessed the speed of optical character recognition using the popular ABBYY Finereader 8.0 package.

In addition, we tested the performance of test systems in the popular Mathematica computer algebra package, the new version of which is able to take advantage of multi-core CPUs.

conclusions

Summing up everything that has been said about the new platform from AMD, we can only admit that the support for DDR2 SDRAM introduced in it is a small evolutionary step forward. Tests show that you shouldn't expect any performance jump from a simple change from DDR SDRAM to DDR2 SDRAM. Moreover, in order to see at least some effect of replacing memory, in tests it is necessary to use the fastest DDR2 SDRAM with a frequency of 800 MHz and minimum timings. The currently widespread DDR2-667 SDRAM may not allow at all to get a performance increase as compared to Socket 939 platforms equipped with DDR400 SDRAM with low latencies.
In conclusion, I would like to add that the appearance of the Socket AM2 platform, working with DDR2 SDRAM, should not be judged as an ordinary event. Despite the fact that at the moment Socket AM2 systems do not have clear and indisputable advantages over the Socket 939 platform, in the future the effect of this transition will become more than clear. Undoubtedly, DDR2 memory is much more promising today. It dynamically increases its frequency and bandwidth, becomes cheaper faster, and, in addition, allows you to create DIMMs of larger capacity. As a result, AMD will undoubtedly benefit from the fact that it has relied on DDR2. Moreover, at a very opportune moment: now no one will scold the manufacturer for such a step, either from the point of view of speed or from the point of view of the price aspect.
However, at the moment AMD is not experiencing real pressure from Intel. Processors from this manufacturer continue to be leaders in almost any application. This is facilitated by the increase in the frequency of the older models of dual-core processors Athlon 64 X2 to 2.6 GHz, and Athlon 64 FX-62 to 2.8 GHz. Of course, there is a danger that the current state of affairs will be reversed with the advent of new Intel processors with the Core microarchitecture. However, it is too early to talk about this.
I must say that after getting acquainted with AMD processors with revision F core, some disappointment remains in my soul. The fact is that the company's engineers once again got off with cosmetic alterations and abandoned deep microarchitectural improvements. It is this attitude of AMD towards improving its own processors that will sooner or later lead to the Athlon 64 family losing the "arms race" to competing processors. Unfortunately, at the moment there is no information about the planned significant alterations in the K8 microarchitecture.