We tested the Bulldozer: FX-8150 and three 990FX motherboards on the test bench (THIS IS ONE)

We know about the page break problem, it will be fixed as soon as possible, until then the article can be read "at the same time"!

Introduction

June 1999, 23 A nice, quiet, simple summer day, but a significant point for a processor company in Sunnyvale. It was around this time that the AMD K7, or AMD Athlon’s seventh generation x86 processors, was introduced. We would only report one data on this: 12 years, 3 months and 19 days. The wrinkled variants of the K7 architecture have been dominant for so long and so much has had to wait for AMD’s next big throw. Apparently, towards the end of this period, the company’s smile wasn’t sincere, but at the same time it is a good example of how long the life of a well-managed processor architecture can be extended.

The glorious past [+]

In recent years, AMD's biggest problem has been that chips with a significantly larger silicon area have to be sold at a lower price than the competition, which is an acceptable solution in the short term, but has a strong impact on expected profits in the longer term. Aware of this, the engineers began developing the Bulldozer microarchitecture roughly 4-5 years ago. Although the floor area of ​​the technologically leading FX CPUs is still relatively large - the official figure is 315 square millimeters - the Intel Sandy Bridge-E chip has far surpassed even this. The transistor density of the chip is quite high, the chip contains just over two billion transistors!

[+]

An interesting development took place during the birth of this article. PC Games Hardware vented that Orochi chip-based server processors contained “only” 1,2 billion transistors. Nothing explains the huge difference, that is, though. AMD originally advertised a bad value for the Bulldozer architecture. In fact, he only reported an approximate value, yet it is obvious that he could not have been so wrong. In recent times, such a well-functioning marketing machine has stalled for the second time. Exact values ​​were not announced, but even at the beginning of the year, news came to light that the new FX processors would also squeeze the Core i7-980X. Even a “random mistake” supports how prepared they were to grind the six-core cannon.

Such was… [+]

… And so it became [+]

On the day of the launch, the company usually launched the new series with a visual video. In addition to head-to-head with Intel’s flagship Core i7-980X and Cinebench R11.5, there’s even a significant advantage over AMD. The rival’s solution scored 5,41 points while the FX-8150 climbed to 5,98 points. The skew is obvious to the open eye. After the hype after the launch subsided, the video became obsolete, but after a little fix, it reappeared on the processor manufacturer’s Youtube channel. The score remained at 5,41, but this is now achieved by the Intel Core i5 2500K.

Technológia

The chips are made on GlobalFoundries ’32nm SHP node. For Bulldozers, the SOI previously introduced has been combined with Intel’s HKMG (High-K Metal Gate), which can help combat leakage current. The architecture is well-designed to achieve high clock speeds (“Speed ​​racer”), making the model range likely to be completely devoid of products below 3 GHz. All the central units in the old language are Black Edition, therefore it is now not specifically marked.

At this point, let’s take a little detour and look at the other side of the coin as well. The fastest quad-core Phenom II processor is ticking at 3,7 GHz, and the 1100T based on the six-core Thuban chip is ticking at 3,3 GHz. In comparison, the 32nm AMD FX-8150 has a near-disappointing base shot and only the 4,2 GHz “level” of the Turbo Core is acceptable, which immediately promises a 10-15% power surplus (no). XbitLabs ventilated a year ago that the Bulldozer was crossing the 3,5 GHz clock, which came together, but in spite of a series of slips. It seems correct to assume that there are still serious problems with the production and output of the new cannon, which has a significant impact on performance.

The second integer only increases the size of the module by 12 percent. [+]

Based on many years of experience, even a basic concept was born, which was based on the following: central units perform fixed point operations at an average rate of over 80 percent. From this, it can be seen that floating-point calculations are much less present in the lives of “centipedes”. In the design, consequently, two integer cores are connected, which have their own first-level cache, but already have to share the second-level cache and the floating-point unit. AMD has named the unit as a module.

One module [+]

According to internal measurements, the second integer basically increases the size of the module to a negligible extent, in contrast, it can ideally cause a performance increase of up to 80%. The part of the primary cache responsible for data is directly connected to the processors (size 16 Kbyte, delay 4 clock), but the 64 Kbyte cache designed to store instructions is already shared among the integrals.

Digging deeper [+]

Based on the test results, the L1 data cache is not only too small, but even slow and this together is not a very good combination. The size of the second-level storage shared within the module is satisfactory, but its latency is high, 25-27 cycles. It is easy to imagine that a larger L1 cache and a faster L2 (12-15 cycles) would improve processor performance by 10-20%.

Not surprisingly, achieving 8MB of L3 is not at the speed of light either (65 cycles). In summary, the Bulldozer cache system will not be the eighth wonder of the world.

In the instruction set maze [+]

Bulldozer currently has the widest set of instruction set support: MMX, SSE, SSE2, SSE3, SSE4A, SSSE3, SSE4.1, SSE4.2, AVX, AES, FMA4, XOP, PCLMULQDQ and of course the 64-bit extension. Of the two novelties (FMA4, XOP), FMA4 is of great importance in the HPC market, and XOP offers a small advantage over multimedia applications. As far as we know, the latest version of x264 already supports the new instruction sets. Outdated 3DNow! support has been discontinued, I think it does not cause many readers sleepless nights.

It is known to use your Intel VT to access x86 virtual memory. IOMMU significantly increases system virtualization performance, however, surprisingly, Intel’s high-end solutions (Core i5-2600K, i7-2600K) do not support this technology and this “black circle” includes current Sandy Bridge E solutions. Again, an extra service compared to direct competition, although its usefulness to the average user is questionable.

Turbo Core in theory [+]

The Turbo Core has also been further developed, working with multiple clock gates and even better adapted to varying degrees of utilization. If all cores are active but floating point units are not currently in use, the Turbo Core 2.0 clock will take effect. The procedure dynamically changes the clock signals of the cores as a function of the load, inactive resources, modules and components inside the module can be disconnected, so there would be no complaint in this area to the front of the housing. Unfortunately, the software side of the thing throws you into the soup thoroughly.

Practical implementation [+]

Windows 7 Scheduler is, to put it mildly, not the most efficient way to allocate tasks because it frequently alternates the allocation of tasks between cores. The next version of the operating system will fix the problem and a fix will be made for this system soon, so - in extreme cases it can be 15-25 percent - we will soon get 2-10% more performance. Another very nice benefit will be that the idle power consumption can be reduced by 4-5 watts because the modules can stay “on savings” longer.

"Don't treat your teeth as a gift" [+]

The "transformation" during Battlefield 3 [+]

Battlefield 3 also shows well how much some optimization helps a processor. In this game, the currently most powerful FX series processor can achieve the performance of the Core i7-2600k.

The FX series processors come with a Socket AM3 + enclosure and are housed in AMD motherboards with a 9-series chipset. Orientation is also facilitated by the color of the socket, which is mostly black. To implement the infinitely sonic Scorpius platform, we need an FX-series processor, a motherboard with a 9-series chipset, and a Radeon HD 6000-series video card. The Bulldozer has a dual-channel DDR1866 memory controller that supports 3 MHz modules.

AMD FX-8150 with a Phenom II X4 970 BE - from above [+]

In conclusion, we would like to add another interesting addition. The fact that the work done by Bulldozer-based processors per clock (instructions per cycle) has decreased somewhat on average compared to its predecessor has caused serious controversy. Some immediately envision the fall of architecture, others list similar examples from the past. In this regard, as always, let us confine ourselves to the facts. Programmers today are increasingly realizing the benefits of multi-core optimization. With an 8-cylinder engine that basically delivers good performance, we rarely think about what it can do with 1 cylinder.

AMD FX-8150 with a Phenom II X4 970 BE - bottom [+]

The example is not the best, but it may shed light on the point. We do not claim that we will make optimal use of eight integer cores densely, but Turbo Core 2.0 targets the highest possible clock (4,2 GHz) in this case. What is only available in the case of K10.5 at the price of “bloody sweat” is considered a “base clock” here. There is also no doubt that the implementation of AVX, FMA and XOP has cost a significant set of transistors. The basics of architecture are used in several segments (server, desktop PC), so this seemed like a mandatory step, but today we see even less of its benefits (especially in a desktop environment).

Lying in bed [+]

Ideally (FMA4 + AVX), the Bulldozer really feels very elementary, delivers surprising performance and puts things in a different light right away. According to measurements from the German HT4U, during the C-Ray 1.1 rendering application, the AMD FX-8150 performs in the same 15 seconds as the Intel Core i7 990X. That’s exactly half the time an AMD Phenom II X6 1100T processor took to do the job. We would note in parentheses that we also weighed the other extreme, Super PI.

Motherboards under Bulldozer

ASUS Sabertooth 990FX

Within the ranks of ASUS, the TUF (The Ultimate Force) family of sheets is perhaps even more special and special than the ROG series. When making TUF sheets, tuning is not the focus (it is also perfectly suitable for standard functions), but we get the best in everything else. The primary goal is reliability and longevity. The first and most important condition for this is the use of quality components and materials.

The components of TUF boards are "Military Class" certified. What it means? Crease-resistant performance in the harshest environments thanks to the use of robust TUF coils, solid-state capacitors and MOSFET components that meet independent partner-verified military-grade specifications. TUF coils, also known as metal alloy coils, are not made of ordinary iron, but of an alloy of various metals, so they can cope with a rated current of up to 25 A, which is 40 percent higher than the load capacity of a conventional component.

And the one-piece housing eliminates the emission of vibrating sounds, resulting in excellent characteristics and durability, even in extreme conditions. Of course, this is only half of the coin, they also need a proper power supply for reliability. This is also done by Digi + VRM for the Sabertooth 990FX, working with 8 + 2 phases. We’ve heard a lot about this, but it’s even complemented by ESP.

ESP Efficient Switching Power provides optimal power supply to key components. The system, which is found exclusively on this motherboard, provides power to the switch not only to the CPU, but also to other key components (video card, LAN, and USB 3.0). ESP significantly increases the efficiency of the system and reduces the heat produced. In addition, ASUS pays special attention to cooling for TUF tiles and uses a special ceramic-containing process.

The ceramic coating of the CeraM! X heat sink coating technology provides 50 percent more surface area for heat dissipation. Innovative ceramics efficiently dissipate heat from the system. The ceramic used instead of the traditional antioxidant layer dissipates more heat due to its microscopic, irregular - and therefore larger - surface. Better cooling always contributes to better system stability. TUF Thermal Radar is responsible for the control.

TUF Thermal Radar monitors the temperature of critical parts of the motherboard in real time and automatically regulates the speed of the fans so that the system maintains excellent stability and does not overheat. There are a number of sensors next to many parts of the motherboard, and it is even possible to monitor them individually. Thermal Radar automatically calculates the ideal fan speeds based on the various parameters specified by users for each component, so everything stays cooler and more durable.

Although we usually do not attach great importance to the matter, electrostatic protection can also "save lives". The Sabertooth 990FX does not ignore this question either. ESD Guards are unique protection of motherboard components against unexpected electrostatic discharges (ESD). Electrostatic discharges happen lightning fast, and their effects are often underestimated. ASUS exclusive anti-charge chip, protection circuits and I/O backplate offer 5 times better protection to extend component life. According to the information in the above paragraphs, it is not surprising that ASUS offers a XNUMX-year warranty on the motherboards found in the TUF family.

This model also measures 30,5 × 24,4 cm using the frames provided by the ATX standard, the color of the printed circuit board is the usual black. The area around the AM3 + socket is quite airy and tidy, the rib system extends beyond the power supply to the 990FX north bridge with the help of a heat pipe. In addition to black, the colors of green and brown are predominant on the sheet, thus providing a kind of military style to the product. We find the usual four DIMMs next to the socket, with DDR3 and dual channel support. The limit is 32 GB in terms of capacity and 1866 MHz in terms of clock speed.

Power is also provided with a 24-pin and an 8-pin power connector on this model. Beneath the RAM slots is the SB950 South Bridge, with an unadulterated TUF rib on it. The same is true for storage storage as with the M5A99X. The AMD SB950 provides native SATA 6 Gb / s ports, six in number, that support RAID 0, 1, 5, 10 modes. A JMicron JMB362 controller has also been added to the board, offering two additional SATA ports, a Power eSATA and an eSATA to its host. Each port is capable of speeds of 3 Gb / s.

The Sabertooth 990FX will not be a bottleneck for its owner either in terms of expansion rails. You get no less than four full-width PCI Express slots, of which black is limited to × 4 mode. It can also be used with three brown video cards, with × 16 / × 16 for two VGA systems and × 16 / × 8 / × 8 mode for three controllers, be it CrossFireX or SLI. We get a piece of a short, × 1 PCI Express, just like we get from the old PCI bus. We won't miss USB connectors either, the SB950 provides 14 2.0 ports (ten of which are on the back and four as pin pins on the PCB). In terms of USB 3.0, the ASMedia controller already mentioned for the M5A99X EVO is back, with four clips on the back and two pins on the PCB. LAN works with a Realtek 8111E, audio with a Realtek ALC 892 8-channel chip.

Sabertooth 990FX back section:

Key features summarized:

• TUF CeraM! X heatsink coating technology - Ceramic coating provides 50 percent more surface area for heat dissipation
• TUF Thermal Radar - Real-time temperature measurement and heat dissipation
• DIGI + VRM
• TUF components (metal alloy coils, capacitors and MOSFET components; military specifications)
• ESP [Efficient Switching Power] - Optimal power supply for key components

Specification table:

ASUS Sabertooth 990FX gallery:

ASUS Crosshair V Formula

Members of the ASUS Crosshair series are the most beautiful dreams for AMD-minded, tuning and gaming-conscious customers. Crosshair V Formula (and Crosshair V Formula ThunderBolt) seeks to carry on this beautiful tradition. The fresh ROG board, of course, got the latest and greatest of everything currently available. AMD 990FX north bridge, SB950 south bridge, AM3 + socket. He is waiting for the Bulldozer with open arms, but is also willing to work with previous models up to the AM3 socket.

In this case, the power supply is provided by Extreme Engine Digi + VRM, with 8 + 2 phases. Needless to say, the design and style of the ATX standard 30,5 × 24,4 cm printed current follows the usual ROG recipe, true to the material and surface of the ribs, the use of color, and this to the unique ROG properties. In addition to the AM3 + socket, a densely laminated black rib runs along the 990FX chip and in the VRM area, with a heat pipe connection.

DDR3 modules can work in two-channel mode in the four DIMM slots, with a total capacity of up to 32 GB and a speed of 2133 MHz. Beneath the RAM sockets is the south bridge, on which a flattened, ROG-designed rib culminates. The advantage of the 990FX North Bridge is also evident in the Crosshair V Formula, for which we need to look at the area of ​​the expansion rails. This model also wears a total of four full-width PCI Express rails.

This allows you to use up to four video cards simultaneously (Quad SLI and Quad CFX), but since the fourth slot can only work in × 4 mode, this can be a limiting factor for the card. The specification therefore officially includes Tri-SLI and 3-way CrossFireX, the three rails can operate in the following modes: × 16 / × 16, or × 16 / × 8 / × 8. The latest Crosshair also comes with a short × 1 PCI Express and a PCI slot. What is certain is that with a well-tuned Bulldozer and a brutal 3D subsystem, this motherboard will have a chance to break records.

Regarding SATA ports, we can practically quote the other two boards: The AMD SB950 provides native SATA 6 Gb / s ports, six in number, which support RAID 0, 1, 5, 10 modes. A JMicron JMB362 controller has also been added to the board, offering two additional SATA ports, a Power eSATA and an eSATA to its host. Each port is capable of speeds of 3 Gb / s. The network controller in the Crosshair V Formula is no longer an Intel solution, but an Intel solution, and the sound generator is also a much more steep copy in the person of the Supreme FX X-FI 2 8-channel controller, with support for many Creative features.

Legend has it that SupremeFX X-Fi 2 endows games with stunning sounds - the dream of every ROG staff member. EAX 5.0 and OpenAL provide realistic, cinematic sounds. The THX TruStudio Pro branding guarantees that games, music and movies sound even better. SupremeFX X-Fi uses 2 gold-plated connectors and high-quality capacitors to offer the best quality.

The SB950 has already been noted to provide only a USB 2.0 port, in this case 12 (8 of which are on the back, 1 is reserved for ROG Connect). Of course, the Crosshair V Formula wouldn’t be able to afford not to carry some of the newer standard either, so came the good old ASMedia controller, offering six USB 3.0 adapters, four of which are on the back.

Here is the back structure and connectors of the Crosshair V Formula:

In addition to what has been described so far, of course there is much more that could be written, for example about the exclusive ROG "features", but since we have visited quite a few ROG models recently, and we have covered these features in detail more than once, we will not do this again now, but summarize.

Key features summarized:

• Three-card NVIDIA® SLI ™ / AMD CrossFireX ™ technology support
• Extreme Engine Digi + - High-performance combination of analog and digital design elements
• SupremeFX X-Fi 2 - Play with realistic, cinematic, surround sound
• GameFirst - Speed ​​with traffic control
• Intel® Gigabit LAN
• UEFI BIOS - Flexible, easy-to-use BIOS interface

Specification table:

ASUS Crosshair V Formula Gallery:

MSI 990FXA-GD80 (ver 2.2)

This minute has finally come: we were able to welcome an AMD MSI motherboard to the editorial office, as well as the manufacturer's most powerful model, the MSI 990FXA-GD80 (ver 2.2). As its name suggests, it is, of course, an AMD 990FX / SB950-based solution, and the GD80 is meant to indicate the highest equipment and best technical implementation. The printed circuit board, which fits into the ranks of the ATX standard, is 30,5 cm long and 24,4 cm wide, so it is no problem to install it in a normal ATX hut. It has now become almost common for manufacturers to build on a mid-range black PCB, which is no different for the 990FXA-GD80. Of course, this doesn’t have too much practical significance, but it gives the eye a nice, sophisticated look, especially with details.

Power is supplied by the standard 24-pin and 8-pin auxiliary power connectors. Around the AM3 + socket - which accommodates AMD FX, Phenom II X6 / X4 / X3 / X2, Athlon II X4 / X3 / X2 and Sempron CPU processors - we find the power section, which is worth mentioning four things. The first is that the board works with a 10 + 2 phase power supply. The second is a technology called Active Phase Switching. It is not difficult to translate, it certainly means active phase switching, which in practice means that the unused phases and the related components are switched off in case of idle, in other words the phase switching is in accordance with the load.

This is not a new thing, we have seen similar things elsewhere, but it is important to emphasize that MSI is not a software solution, the control is fully automatic, hardware based, the thing is done by a special IC chip. This is not useless, as it increases efficiency and, as consumption decreases with use, it also results in long-term savings (the blue LED bar next to the socket on the motherboard informs about the current load and the number of phases in use).

The third thing is DrMOS. Anyone who has been a bit in the picture with MSI motherboards lately has definitely met DrMOS, as the company has been using this solution on their motherboards for a long time, all the way back to P45. DrMOS stands for Driver-MOSFET, a power supply component.

The essence of the technology is that the driver IC required for control and the upper and lower MOSFETs are housed in a single chip, a case. One of its important virtues is the increase in switching frequency, which, according to the manufacturer's data, means four times the speed compared to a conventional MOSFET, which results in a significant improvement in the stability in voltage regulation. It retains its stability and efficiency not only under general conditions, but also in the case of hard tuning. The fourth thing is Military Class II, which deserves a little more words and applies not only to the PWM area, but to the entire motherboard. A few paragraphs up, you could also read about a motherboard that was built on components that also met military standards. In the case of MSI, it is practically the same. Military Class II, as its name suggests, is already the second revision, but the point is in the details, that is, in the parts.

Hi-c CAP

The soul of Hi-c capacitors is tantalum, which is the 73rd element of the periodic table, and is a rather rare, off-white, shiny metal. Its melting point is very high, 3017 ° C. Due to its high chemical resistance, chemical devices and electrodes are made of tantalum. It is used to make dental drills, analytical balance weights, cathodes of X-ray tubes, fountain pen tips. Because it fits well with tissues, it is used to make aids that remain inside the body, such as supporting bones. It is also widely used mainly for capacitors in mobile phones and other electronic equipment, as shown in the attached figure. As a result, the Hi-c capacitor is estimated to have an eight times longer life than a conventional solid electrolyte capacitor.

SFC

SFC (Super Ferrit Choke) covers a specially designed iron core coil. The coil contains an iron core whose permeability has been significantly increased. Compared to conventional coils, the SFC works at a temperature 35 ° C lower. Of course, this also has a small positive effect on heat production and stability.

Solid CAP

Solid CAP, the solid electrolytic capacitor, is the starting point for the Military Class series. These models no longer include conventional electrolytic capacitors, all conduits are at least Solid CAP, but where needed, Hi-c capacitors are placed there. This way we no longer have to worry about exploding, bulging capacitors, these types of conductors fail in a negligible number, with a nominal life of 10 years.

From these components, Military Class II is formed, and in the package we also find a certificate that guarantees the authenticity of these parameters. A thick, nickel-plated aluminum rib has been placed in the PWM area, and the connection is provided by a thick, flattened copper heat pipe to the rib on the north bridge, which is also nickel-plated. Of course, there was a bit of attention to the ribs as well, we can read Military Class II and OC Genie II on a 1-1 blue plate, which are two defining features of today’s MSI cards. The rib is the same material and style on the south side of the SB950 south bridge, of course its shape is flattened so as not to prevent the installation of long video cards. Here you will find the MSI brand.

As is common with ATX standard AM3 + boards, this model also features four DIMM sockets that, of course, accept DDR3 modules up to 2133 MHz with a maximum capacity of 32 GB. In use, dual-channel is supported. The colors are also distinguished by the colors, one channel is marked in blue and the other in black.

Let's look at the SATA section. At the edge of the PCB, we find six ports rotated, all coming from the SB950 South Bridge and supporting the latest 3.0 standard. If you fly across the opposite edge of the printed circuit board with a light gesture, you may also come across the gigabit LAN controller, which is a Realtek 8111E instance, as we have seen many times, a popular chip. You can also see a THX sticker in the corner, below which is the audio controller, which is a Realtek ALC 892 chip, and this is not our first encounter. It provides up to 8 channels with jack detection, Azalia 1.0 and Vista Premium support.

We are not left without things worth mentioning in the lower part of the motherboard, the most characteristic of which are the three physical control buttons. The functions of "reset" and "power" are clear, but OC Genie hides excitement. MSI 990FXA-GD80 has OC Genie II support. There are two ways to achieve this. One is the already mentioned button, pressing which results in a one-time and immediate tuning. Therefore, those who do not feel enough routine and courage for manual tuning can safely ask for the help of the OC Genie button. The other way is, of course, access through the BIOS.

Speaking of BIOS, I'd like to interrupt the train of thought here. MSI also uses UEFI BIOS, i.e. the entire environment is graphics-based. MSI shaped this a little on its own face and gave its work the name Click BIOS. This model is already equipped with Click BIOS II. This resulted in a very simple and clear, transparent graphic interface, which can be controlled not only with a mouse, but also with a touch screen. The interface looks almost the same as under Windows, the individual operations can be performed even by dragging the mouse, using the "drag & drop" method. Of course, we get full access to all the capabilities of the motherboard, including monitoring and high-level tuning. OC Genie II functions can be accessed from Click BIOS II.

Below, in the remaining area, we find the pin terminals, with a focus on USB. The first is marked in red, which means that the buckles connected to this terminal have a Super Charger function. You have a good idea of ​​what this means, quick and easy charging of our devices, even when the machine is switched off. The other (black) is a simple USB 2.0 port.

This model has four full-width PCI Express rails, so even the 4-way CFX can be installed. The first two slots can operate in ×16 mode, the third in ×8 mode, and the fourth in ×4 mode. In addition to the four "long" rails, we also get two short ones with ×1 speed, as well as a traditional PCI rail. A diagnostic display is located under the rib of the southern bridge, which consists of two seven-segment displays. Its task is usual, i.e. communicating possible errors with the help of error codes. Most of the motherboard has been completed, only the back section remains to be described:

The line is opened by two PS/2 connectors for the mouse and keyboard. These are followed by a small microswitch, which tuners will be happy to receive, because it is a "CMOS Clear" button. This is followed by coaxial and optical digital outputs. The next column starts with a FireWire, then two USB 2.0 ports can be observed below and next to it, the bottom two of which can also be used as eSATA ports. The next two columns are very similar, as they both contain four USB sockets, with an RJ45 gigabit ethernet port on the top of the first. The dog is buried with the USBs, because the black ones are only 2.0, while the blue ones are 3.0. The line is completed - as always - by the jack analog audio connectors.

Key features summarized:

• It builds on the latest AMD 990FX chipset
• Outstanding stability thanks to Military Class II
• OC Genie II: Overdrive in the simplest way: significant extra power at the touch of a button
• Support for 3-way SLI and 4-way CrossFireX
• Support for USB 3.0 and SATA 6Gb / s standards
• THX TruStudio PRO: a provider of quality sound

Specification table:

MSI 990FXA-GD80 gallery:

Test programs, test configurations, consumption and warm-up

Applications used during the measurements:

• CPU-Z 1.58-1.59
• AIDA64 2.00.1720 beta
• Sisoftware Sandra 2011.11.17.84/86/XNUMX Engineer Standard xXNUMX
• TrueCrypt 7.1
• Fritz Chess Benchmark 12
• Winrar 4.10 beta 3 x64
• 7-ZIP 7.22 beta x64
• Super Pi mod 1.5XS
• WPrime 2.04
• Cinebench R10 x64
• Cinebench R11.5 x64
• POV-Ray RC3 x64
• Luxmark 1.0
• Photoshop CS4 11.0.1
• X264 HD Benchmark 4.0
• dBpoweramp Music Converter
• MediaEspresso 6.5.2119_41281
• 3DMark Vantage 1.0.2
• 3DMark11 1.0
• Battlefield 3 + FRAPS 3.4.7 build 13809
• Crysis 2 1.9 + Adrenaline Crysis 2 benchmark tool 1.0.1.13
• F1 2011
• Far Cry 2 + Far Cry 2 benchmark tool 1.0.0.1

Test configurations features and settings

AMD FX-8150 + Crosshair V Formula:

AMD FX-8150 unloaded, Super Pi-loaded (single-threaded) and Fritz-loaded (multi-threaded), and RAM settings

AMD FX-8150 + Crosshair V Formula + tuning:

CPU and RAM information

To the best of our knowledge, a single FX-8150 CPU has arrived for domestic media, and this has been touring various editorial offices. We would have been relatively enthusiastic about tuning if we hadn’t gotten a glimpse of the overcrowding experienced at one of the competitions. There, things got to 4600 MHz. We also browsed through the experiences of several foreign sites, accordingly, we turned off the power management features and the TDP limiter, but unfortunately we couldn’t get any further, in fact. During the tuning we only increased the multiplier, but even so we could not make the system stable above 4515 MHz. Surprisingly, this too was only achieved by not increasing the voltage but decreasing it, up to 1,375 V in the BIOS. Even though we tried higher values ​​up to 1,41 V, the higher operating voltage always only resulted in instability. Let us add that the FX-8150 we visited is an early copy, we are confident that those on store shelves will have more favorable features.

AMD FX-8150 + Sabertooth 990FX:

CPU and RAM information

AMD FX-8150 + MSI 990FXA-GD80:

CPU and RAM information

AMD Phenom II X4 970 BE + Sabertooth 990FX:

AMD Phenom II X4 970 BE unloaded and loaded, and RAM settings

Intel Core i5-2500K + GIGABYTE Z68X-UD7-B3:

Intel Core i5-2500K unloaded, Super Pi loaded (single-threaded) and Fritz-loaded (multi-threaded), and RAM settings

Intel Core i5-2600K + GIGABYTE Z68X-UD7-B3:

Intel Core i7-2600K unloaded, Super Pi loaded (single-threaded) and Fritz-loaded (multi-threaded), and RAM settings

Intel Core i7-980X + GIGABYTE EX58-UD4P:

Intel Core i7-980X unloaded, Super Pi loaded (single-threaded) and Fritz loaded (multi-threaded), and RAM settings

Consumption:

Let's see the "dreaded" consumption! If we look at the numbers, we can see that the FX-8150 can't be called horroristic when it comes to hunger at its base clock, with power management functions on its side. In an unloaded state, it asked for the least, and even during Blu-ray playback it only consumed a few watts more than the two Sandy Bridge CPUs. During AIDA64, it showed more protein in its teeth and consumed the most energy among the field members, with a large difference compared to the Sandy Bridge models. In the case of Furmark, the situation was not so dire: although the FX-8150 consumed the most after the Phenom, the difference compared to the Sandy Bridges was 15-25 watts. It is better not to detail the tuned consumption data, we can clearly see what happens when we unchain the monster (turn off restrictions and energy management) and unleash its appetite.

Warming:

The heat generation characteristics of the AMD FX-8150 were tested in the ASUS Crosshair V Formula sheet:

In the first round, the monitoring was performed with basic settings, leaving all power management functions switched on. The FX-8150 CPU was loaded using AIDA64 for 30 minutes and then allowed to cool. During loading, the total temperature of the processor has risen to 59 degrees Celsius, while that of the cores has risen to 48 degrees Celsius, which does not seem unfriendly, but FX processors are already known not to like high temperatures, up to 60 degrees can be said to work under the right conditions . I managed to keep this, but with a Scythe Ninja 2 Rev B, which certainly can't be said to be an average and cheap cooling, so the question arises as to what happens if we snap a cooler of a few thousand forints onto the FX-8150. Well, we don’t know that, but it’s not out of the question that there could be a problem with a longer load on every core.

Seeing this, we were a little afraid to put the load on the CPU in the tuned state, as we knew we wouldn’t fit into the limit. During loading, the temperature crawled up to 79, both in terms of total value and individual seeds. This certainly slips far from the recommended value, so if our cooling is inadequate, let’s give up permanent tuning rather than torture our unit. The problem is that the Ninja 2 Rev B is not a weak model, although it is no longer the pinnacle, the latest products are not able to beat it by a huge difference either, so we have come to the conclusion that if we want an FX-8150 with lasting tuning to use, we would most likely need water cooling. AMD recognized this fact much earlier than we did, which is probably why it decided to make the FX-8150 available in a package that includes a factory water cooler. The cheaper edition includes a heat tube air cooler, so don’t cherish big dreams about OC.

test Results

Memory management:

AMD FX-8150 + ASUS Crosshair V Formula (basic) AMD FX-8150 + ASUS Crosshair V Formula (tuning)

AMD FX-8150 + ASUS Sabertooth 990FX AMD FX-8150 + MSI 990FXA-GD80

AMD Phenom II X4 970 BE + ASUS Sabertooth 990FX Intel Core i5-2500K + GIGABYTE Z68X-UD7-B3

Intel Core i7-2600K + GIGABYTE Z68X-UD7-B3 Intel Core i7-980X + GIGABYTE EX58-UD4P

AIDA64 read speed test (read test suite):

AMD FX-8150 + ASUS Crosshair V Formula

AMD FX-8150 + ASUS Sabertooth 990FX

AMD FX-8150 + MSI 990FXA-GD80

Synthetic measurements:

In describing the results, we will mainly compare the FX-8150 to the two Sandy Bridge models, as they are priced in about the range that the FX-8150 comes in. During the built-in measurements of AIDA64, the shape of the FX changes greatly for the various modules. Under Queen, you only have a chance against 2500K, not even 2600K when tuned. Photoworxx has no chance of tightening the 2500K and 2600K, but surprisingly, the FX-980 manages to handle the 8150X. During the Hash module, the Bulldozer felt spectacularly good and beat all his opponents with confidence, with a surprisingly large difference. Under VP8, the situation is no longer so rosy, the 2500K has been approached, but the 2600K and 980X are stepping up. Tuning, on the other hand, has swung a lot, and AMD’s new savior has taken the lead. Under FPU Julia, even the strongest Bulldozer doesn’t excel so much, he only manages to deal with Phenom.

Like AIDA2011, Sisoftware Sandra 64 is a good example of the undulating performance of the Bulldozer architecture and the FX-8150. In one measurement he lags ugly, in the other he catches up, while in a third he is able to beat the complete field. Tuning could also result in significant percentages of values.

Since the Bulldozer’s strength lies in multi-threaded processing, we didn’t expect a big spark on the Super Pi, this progi was somehow never a favorite of AMD processors. Our conjecture was confirmed in the worst case scenario. The FX-8150 even got out of Phenom, with Intel CPUs moving in a different dimension. So it’s better to forget about Super Pi in relation to the Bulldozer.

WPrime is also a dry, synthetic test program, but unlike Super Pi, it can take advantage of more cores. Unfortunately, it doesn't seem to be enough, as the FX-8150 was included in this test program as well. The Phenom II was much better able to keep up with Intel units than the new architecture.

The Fritz Chess Benchmark (Deep Fritz) is now capable of handling up to 12 threads, meaning the 980X could put all its power into this measurement. You can see that, as he took the lead toweringly, but this test was already to the liking of the AMD FX. We can’t say that we got the right result for the eight cores, as the 2500K was just beaten and the 2600K was lagging behind (tuned), but it was clear that the FX-8150 likes it when all the eight cores may be in use.

Compression and encoding:

We used 7-ZIP for the first time in our measurements, maybe it was time. We have installed the latest beta version, hoping the differences between the amount of cores (and threads) will be nicely visible. That's how it happened: not surprisingly, the 980X took the lead, while the FX-8150 was able to capture the 2600K and tuned up significantly.

We deployed the latest version of Winrar available at the time of the measurement, relying on the "Bulldozer support". This step was not in vain, as the FX-8150 practically outclassed the field, beating both the 2600K and 980X by a large margin. Encouraging signs for the future.

DBpoweramp is also a newly deployed application, a simple and fast coding program. During the measurement, we captured a 93 MB FLAC file that was transcoded to MP3. At the end of the process, the program displayed the time required for the transcoding, and we recorded this and engraved it in the graph. This brought out the weaker form of the Bulldozer and came to the end of the field, with Intel CPUs again leading the way with a half-circle lead.

MediaEspresso is already a routine application in our tests. It is advantageous because in addition to CPU power, it can also be used with GPU acceleration and knows a lot of formats. We transcoded the 1080p raw material into four formats with pure CPU power. The results show that the strength of the FX-8150 depends largely on the particular encoding format, but Intel processors, more specifically 2500K, have only been squeezed or defeated at m2ts and h.264.

The X264 HD Benchmark, as the name implies, specializes specifically in this format, a quick and easy test program that runs on the command line and reports the result when coding is complete. FPS values ​​were plotted on the graph. 1 pass brought the superiority of Intel CPUs, while 2 passes the FX-8150 managed to beat 2500K and reach 2600K.

TrueCrpyt can be used to test the speed of operation associated with AES encryption. The Phenom II does not yet have hardware AES support, so we can see what. The other CPUs, on the other hand, have the capability, and the FX-8150 is now performing quite well. It clearly beats the 2500K, and manages to approach the 2600K nicely, and it beats it hard with tuning. The 980X is the absolute king.

Rendering:

Cinebench R10 is an old engine, but it's still worth using because you can see how much the result achieved with one core matters when all cores and threads do the rendering. It's good to see the theory becoming practice in relation to Bulldozer, according to which "core-to-core" is somewhat weaker than the previous generation, not to mention the Sandy Bridge architecture. In the second measurement, the eight cores come to life, and things immediately paint a different picture, but 2500K can still only be approached. On the other hand, the tuning produced serious results, the 2500K and 2600K were also beaten, only the 980X could overtake them.

Cinebench 11.5 is the latest version of the previous application, here you can only measure using all cores+threads, but this program is nowadays considered a serious value measure in relation to multi-threaded instruction processing. In the introduction to this program, AMD's "graphics joke" about the FX-8150 and 980X came up, but now we can see the harsh reality. It doesn't have much to do with the 980X, the 2500K is beaten by a small margin, but don't forget that the FX-8150 works with twice as many cores as the 2500K. Accordingly, the 2600K can no longer cope. You can also see that a possible new stepping, which could be sent into battle with a much higher clock signal, would help the situation of the Bulldozer a lot, because the patient accelerated a lot again as a result of the tuning.

We rested the POV-Ray for a while, now re-elevated it back into the test apps for the sake of the Bulldozer. The latest version, 3.7 RC3, was used, and the time required for rendering and the score obtained were also recorded. It can be seen that the FX-8150 thanked for the confidence this time, stepping down the Phenom, beating the 2500K with confidence, and now it has managed to beat the 2600K, if only by a hair. The 980X failed to put on the glove, but that’s no big surprise.

LuxMark is also the first software to measure rendering capabilities, even in GPU-accelerated mode. We now used only the processor power and recorded the scores we received. We expected a little more from the FX-8150, the four-core 2500K was almost mandatory to beat, but the 2600K and 980X were significantly behind.

We close this section with Photoshop, in which we ran a script that draws a nice calendar on the screen by the end of the operation. The time required for execution was measured, so the smaller value is the better in the graph. So far, AMD hasn’t liked Photoshop, and it doesn’t seem to be going any further (the truth is, we’ve tried CS5 as well, but we’ve gotten even worse results). Intel CPUs compete one class up, and the Bulldozer is just dealing with Phenom.

3D and games:

Finally, follow the two 3DMark and games! We were curious to see what forces the current king of Bulldozer architecture could mobilize in this area. Under Vantage, the situation is a bit controversial, as the FX-8150 lags behind Intel CPUs in terms of performance score and was only able to outperform the Phenom with the MSI GeFroce 560 Ti, while the CPU score shows that the FX represents more power at 2500K. The tuning threw in the result pretty nicely, the 2600K and 980X still couldn't be caught.

3DMark11 is already a much newer software, with DX11 support, and appears to be much more Bulldozer-friendly than its predecessor. During the Performance run, I managed to depress 2500K, only the 2600K was only a hair's breadth behind the FX, and with tuning it took the lead. The Xtreme score lay even better for AMD’s new miracle, which has put all its opponents behind it overall.

Let's see what Crysis 2 shows with the DX11 patch and the latest measurement software! Moving towards higher resolutions will be more and more limited, but to see the difference at the two lower resolutions, and clearly in favor of Intel processors, although the difference can only be called significant at 1024 × 768. Tuning did not matter much during the measurement.

During Battlefield 3, unfortunately, we were only able to install FRAPS measurements, also by doing the same route at the beginning of a given course, for approximately the same amount of time. Nevertheless, the measurement may not be 100% accurate, but perhaps the point is visible. At 1024 × 768, even the Core CPUs drive, but the FX-8150 isn’t far from them, and then switching to 1280 × 1024, the Bulldozer also hits the two Sandy Bridges. The two higher resolutions are moving dangerously towards the limit, but it looks good that the FX-8150 is not lagging behind the competition, in fact, it manages to catch up with the 2500K and 2600K and reach the 980X. Tuning had little effect on FPS values.

Unfortunately, F1 2011 already paints a much more negative image of the FX-8150 than the previous two titles. What you have seen does not need to be explained much, Intel CPUs in all resolutions are badly outperforming AMD units. Here, the tuning at least resulted in a few FPS, but this time only the Phenom II was suppressed, and not with much confidence either.

We arrived at our last graph with the results of an old acquaintance, Far Cry 2. FPS values ​​followed the same trend as we saw at F1 2011, meaning the numbers prove a total victory for Intel models, it would be better to forget the performance of the FX-8150, but let’s not forget that it’s a very old engine, thus, a maximum of one or two of the eight cores could be in use, and even Turbo Core could not help with this.

Summary, opinion

Let’s say a few words about the AM3 + motherboards in the test. Only the MSI 990FXA-GD80 could have been unknown, as the other two ASUS models could have flashed their capabilities in a previous test, though only with the Phenom II X4 970 BE processor. Each of the three motherboards represents three different styles.

Az ASUS Crosshair V Formula ROG sheet, so no question about when he was born. For play and master tuning. It's no coincidence that DDR3 tuning and FX-8150 also set a world record for overdrive, with The CPU clock has been upgraded to 8544 MHz. In addition to the Crosshair V Formula, we get the standard, thickly filled box with the usual ROG accessories. The price of the motherboard is also adjusted to this, which is currently about 62 ft. As always, we recommend this ROG sheet to those who are striving for the maximum, but it is also important to be able to squeeze serious tuning out of their system. Overall, the ASUS CrossHair V Formula, as last time, is now very much I liked it.

Az ASUS Sabertooth 990FX a different kind of plank. Of course, tuning is not an obstacle in this either, but this model has been invented for above-average stable operation and long life. This is evidenced by the TUF (The Ultimate Force) components (capacitors, coils, MOSFETs), the special ceramic-coated cooling, and the 5-year manufacturer's warranty. In addition, the price of this model is quite friendly compared to its knowledge, it can be taken home around 40 ft, which is not a bad offer, so overall, with this price / performance indicator, Sabertooth deserves to be recommended be.

ASUS Sabertooth 990FX

Az MSI 990FXA-GD80 this is his first visit to us and we were pleased with his performance. He did his job stably during the tests, but he did well with the FX-8150. This board is close in spirit to the ASUS Sabertooth 990FX. As for the former, the TUF component is here the Military Class II. Hi-C capacitors, DrMOS, SFC, these components also meet the criteria of military standards, thanks to the stability and service life offered by the MSI 990FXA-GD80, which, incidentally, represents the top category in equipment. The purchase price of the motherboard for 57 ft, on the other hand, is not small, but if you want to own the best and best-equipped MSI AM000 + board, you have to get that amount from the piggy bank. The MSI 3FXA comes with a 990-year manufacturer's warranty. The paper is not far from the recommended rating, but due to its slightly high purchase price, the I liked it receives a rating.

MSI 990FXA-GD80

Thoughts on the Bulldozer and FX-8150

This time we do not summarize by program, the reason for which is the fluctuation that is currently characteristic of the Bulldozer architecture and the FX-8150. Even within a test, it can show completely different performance regardless of the given area. During the measurements, it became clear that it is important for the FX-8150 to have as many modules as possible, i.e. as many cores as possible, because this is where its power lies. Despite the maximum Turbo Core clock speed of 4200 MHz during single or dual-thread processing, it does not seem to help AMD's new savior. What it already does quite well is compression, and in some cases rendering and coding, which is no wonder, since it is in these areas that we find most of the programs capable of pushing eight cores. Often, however, even among these applications, only the i5-2500K succeeds, which, let's face it, is not a steely performance with twice as many cores on board. During games, the situation is not terrible with the latest, most modern engines (which work with at least four cores), but with the previous engines, the FX gets in trouble. The same can be true for other programs, each new program version can make the given application almost "Bulldozer-ready", if we only think of Winrar 4.10 beta 3, for example.

Factory air cooling for FX-8150 - source: legitreviews

But what about prices? Is it worth it now or is it not worth it? Due to these manufacturing difficulties, very few FX processors are currently rolling off the production line, so there is a general shortage of stock in the FX series with a little exaggeration. Probably this also imprints on prices, but of course the dollar and its exchange rate are not good for us Hungarians either. The recommended end-user price of the FX-8150 is $ 270, which should be around $ 60 at home. Let’s stay at $ 000 first. The Core i270-5K’s current tag shows $ 2500-205, roughly $ 213 cheaper than the most powerful FX. The Core i60-7K costs $ 2600-294, which is only $ 304 more expensive and can be considered much more powerful in performance. In our opinion, the purchase price of the FX-30 is somewhat exaggerated, based on its current performance, somewhere around 8150K, around $ 2500, the amount of money asked for it would be realistic.

Factory water cooling for FX-8150 - source: hothardware

Remember, Sandy Bridge models are often not only faster, but consume less and produce less heat, so cooling them is much less of a task than keeping an FX-8150 CPU cool, thanks to a 60-degree ceiling as well. Let's look at the brittle Hungarian prices! A Core i5-2500K in the polar vegetable can be taken for 55 ft, while an i000-7K is needed for about 2600 ft. In the few places where the FX-79 is also available from stock, it can be struck around 000 ft. Interestingly, the six-core Phenom II X8150 70T is already available for 000 ft, so it’s a good buy for certain tasks today. In light of what has been said, we would not be able to recommend the FX-6 to any of its competitors based on its current price and shape. In comparison, to hear news that the low stock that is available from it is immediately caught abroad, so despite the lower-than-expected performance, there is a lot of interest in the new family.

Is there a more beautiful still life? [+]

Our feeling is that the Bulldozer is a big promise and has remained so for the time being. The tuning highlighted that at significantly higher clocks, the Bulldozer would often be able to do much more, not by chance, as the architecture was designed in this spirit from the outset. However, this is all in vain if the current manufacturing technology and the power associated with the current stepping does not allow a clock signal higher than 3600 MHz for eight cores.

Right now, the best thing we can do is continue to travel with DeLorean and jump forward a few months and then a year. It wouldn’t come as a huge surprise if the first B3 stepping processor, the AMD FX-8170, appeared in the first months of next year. According to preliminary news, this model would mean a performance increase of nearly 10%. So this will be the first, but not the last, step. Already in the spring, an FX series processor (FX-8190?) Could appear that can cross the 4 GHz clock without Turbo Core. It is also clear to the naked eye that the Bulldozer is already capable of high clock speeds, only the power consumption of the top models does not allow it yet. There is one more spare in AMD's hands.

Jewelry or jewelry? [+]

In this segment - if not amid cloudless joy - but the 140-watt TDP is still acceptable. An example of this can be found in both Phenom series. The next stop on the imaginary time travel is the summer of 2012, when Vishera CPUs with the Piledriver module may arrive. According to information obtained by DonanimHaber, we can expect 10% better x86 performance from the novelty. The long-awaited Windows 8 can be completed in late summer and early autumn. It won’t work a miracle, obviously, but we wouldn’t reject a few percent (2-10) acceleration as a gift.

The situation in Faramuci is that while the performance and introduction of the AMD FX-8150 is, to put it mildly, not overwhelming, the Bulldozer architecture seems to be a good foundation because of the examples above. We are confident that in 1 year, AMD’s current flagship will deliver orders of magnitude higher performance than the current FX cannon. However, despite all the negative features of the FX-8150 I liked it.

AMD Bulldozer architecture and AMD FX-8150 processor

The MSI 990FXA-GD80 motherboard is Domestic representation of MSI, while the ASUS Sabertooth 990FX and Crosshair V Formula motherboards are Domestic representation of ASUS provided. The AMD FX-8150 is AMDcame from. The Core i7-980X and the Core i5-2500K are Intel, while the Core i7-2600K Over @ locker886 provided for measurements. Thank you all!

The article was written by: Zoltán Mihics (med1on) and Gábor Pintér (gabi123)

Thanks to the following sponsors for our permanent test components in this article:

• Intel Core i5-2500K - Intel
• GIGABYTE Z68X-UD7-B3 + GIGABYTE Representation in Hungary
• Kingston HyperX T1 DDR3 2133 MHz 4 x 2 GB Kit - Kingston Representation in Hungary
• Kingston SSD Now V Series 64 GB SATA2 (SNV425-S2 / 64) - Kingston Representation in Hungary
• Xigmatek NRP-HC1501 - PC Gear
• ASUS 24T1E TV Monitor - ASUS Representation in Hungary