Korean researchers have developed a method to grow sub-nanometer semiconductor logic circuits. The team from the Institute for Basic Science (IBS) claims that they grew 1D metallic materials with a width of less than 1nm and developed them into 2D circuits. In effect, the 1D metals acted as a gate electrode of an ultra-miniaturized 2D transistor.

The headlining technology could be very important, as it exceeds predictions outlined on the International Roadmap for Devices and Systems (IRDS) by the IEEE. While the IEEE predicted semiconductor node technology to reach around 0.5 nm by 2037 with transistor lengths of 12nm, the IBS researchers demonstrated a 1D MTB (mirror twin boundary) gate can be as small as 3.9nm.

It goes almost without question that creating ultra-miniaturized transistor devices based on 2D semiconductors comes with significant technical challenges. The scientists note that in traditional semiconductor fabrication processes, reducing the gate length below a few nanometers is limited by lithography resolution. To solve this issue the Korean scientists used the mirror twin boundary (MTB) of molybdenum disulfide (MoS₂) as a gate electrode, achieving a width of just 0.4nm.

In a blog post, which is supported by a Nature Technology paper publication, the researchers explain that “the 1D MTB metallic phase was achieved by controlling the crystal structure of the existing 2D semiconductor at the atomic level, transforming it into a 1D MTB.” This is claimed to represent a “significant breakthrough,” for both materials science and semiconductor tech. The artificial control of crystal structures to synthesize the materials is key to this advance.

Interestingly, the 1D MTB-based transistor also boasts some inherent benefits compared to tech such as FinFET or GAA. According to the researchers, their new transistors “minimize parasitic capacitance due to its simple structure and extremely narrow gate width,” resulting in greater stability.

Hopefully, this announcement won’t be just a blip on the radar. Director JO Moon-Ho commented, thinks not, asserting that the 1D MTB-based transistor is “expected to become a key technology for developing various low-power, high-performance electronic devices in the future.”

Intel has proclaimed that “the Wi-Fi future is here” after successfully demonstrating Wi-Fi 7 technology with partner Broadcom. In the over-the-air demos, a Wi-Fi 7 connection between an Intel Core-powered laptop and a Broadcom access point achieved a steady 5 Gbps transfer speed. This is a worthwhile upgrade versus Wi-Fi 6, which maxes out at 1 Gbps.

As the Intel news blog highlights, industry collaboration is essential to ensure the widespread adoption of Wi-Fi 7 and its practical benefits. Having one Wi-Fi 7 in one category of devices but no other is pretty useless.

Broadcom chimed in to assert that “the ecosystem is ready.” It highlighted the value of Wi-Fi 7 goes beyond simple speed increases to deliver greater network capacity and the benefits of much lower latencies. The reduced latencies will be invaluable for immersive experiences over Wi-Fi, with the most obvious applications being online gaming and reduced latency wireless VR headsets. For media streamers, it also opens up the eye-popping possibility of UHD 16K streaming (or more people on the network streaming 4K or 8K videos). Last but not least, the newest Wi-Fi standard could help reduce wireless network traffic jams in busy homes and offices.

The underlying technologies behind Wi-Fi 7 include wider 320 MHz channels in unlicensed 6GHz spectrum and higher order 4K QAM data modulation. Intel’s blog also outlined the importance of multi-link operation and improved channel utilization efficiency.

Video: demo compares Wi-Fi 6 on the 5 GHz band, vs. Wi-Fi 6E on the 6 GHz band, vs. Wi-Fi 7 on the 6 GHz band.

These are still early days for Wi-Fi 7, as it isn’t expected to be certified until next year, with products bearing the Wi-Fi 7 logo perhaps not appearing until H2 2023. However, last month an Intel exec was quoted as saying Wi-Fi 7 will start to be “installed in PC products such as laptops by 2024.” As one (code)name is never enough, Wi-Fi 7 devices will also be referenced as featuring the “Wi-Fi 802.11be” connectivity protocol.

A positive from the lengthy sounding rollout timescale is that Wi-Fi 7 speeds could be further enhanced in the interim. Hopefully, Wi-Fi 7 will catch on quickly, inspired by the attractive transfer speeds and latency benefits. It will also be welcomed if Wi-Fi 7 device rollout isn’t hampered by the component shortage problems, which slowed Wi-Fi 6E device availability.

Quantum computing researchers working for Google have reportedly built a CMOS integrated circuit (IC) that will work at cryogenic temperatures. The chip is intended to replace the 168 coaxial cables that are currently needed to power-up the company’s 72-qubit quantum computer.

From Cables to Chips

As the number of qubits for quantum computers increases, the number of circuits connecting those qubits with one another also has to increase. There are only 72-qubits in Google’s most powerful quantum computer so far, and it already needs more than twice as many cables.

Google believes that the quantum computers that will solve some of the world’s largest problems in the future will need millions of qubits. Suffice to say that it wouldn’t be practical for quantum computers to keep increasing the number of cables they use to connect all of those qubits.

In order to make quantum computers scale up more easily, all of the components will have to work at cryogenic temperatures and that includes the circuits between the qubits. In the first phase of transitioning from cables to a quantum computer that works entirely on a chip, Google’s team was able to design a high-performance, low-power pulse modulator as an IC to program the qubits at cryogenic temperatures.

Controlling Heat Dissipation

Joseph Bardin, a University of Massachusetts at Amherst associate professor of electrical and computer engineering, who has helped design the cryogenic IC, said that the biggest challenge with trying to move from cables to an IC is the heat dissipation.

Although the qubits operate at a temperature of 10 milikelvin, the control circuits are were operating in the range of four K. However, a CMOS transistors generally operate at a temperature that is 200 degrees higher and start to behave differently when operated at such low temperature.

The other issue was that even four K was too high of a temperature compared to the milikelvin temperature range required for quantum computer circuitry. Therefore, Bardin and Google’s team had to ultimately come up with an IC design that used only 2mW in testing.

Bardin also said that there are two other components of a quantum computer that need to be cryogenically integrated to help quantum computers scale up to millions of qubits -- the readout system, as well as another component that controls a different aspect of the pulse modulator.

Intel and IBM, two other major companies that are developing quantum computers, are also said to be working towards the same goal of putting quantum computers on a chip.

Visbit CEO Changyin (CY) Zhou demonstrated his company’s View-Optimized Streaming (VVOS) technology at the IEEE VR Summit by transmitting a 12K 360-degree VR video seamlessly over a consumer Wi-Fi connection and a Samsung Gear VR HMD.

As virtual reality HMDs grow more and more popular, so will the demand for high-quality content. Many people, especially those who aren’t interested in gaming, will in turn look towards 360-degree VR video for their content desires. There are ways to create high-resolution 360-degree content--several companies make super high-resolution 360-degree cameras, and it’s not hard to find a camera rig that supports an array of 4K GoPro action cameras.  Creation isn’t the problem. Distributing all of that high-resolution 360-degree content—especially stereo 360-degree video—is the hard part.

Increasing the available bandwidth is the obvious solution for streaming high-resolution content, but that’s easier said than done. Advancements are being made in networking technology, but those solutions won’t be mainstream for some time. Visbit is taking a different approach to the problem. The company introduced VVOS in December 2016, and the company told us that VVOS makes it possible to stream high-quality 4K and higher resolution 360-degree content.

Apparently, Visbit wasn’t kidding about streaming content with higher resolution than 4K.

“Today, 360-degree videos are mostly streamed at 1080p and occasionally at 4K. You rarely hear of 6K video streaming, so 12K, a resolution that is very close to the acuity of 20/20 human vision, is unheard of, especially when streaming on today’s regular Wi-Fi and LTE conditions,” said Visbit Co-Founder and CEO Dr. Changyin (CY) Zhou.

Visbit created a 12K x 6K video which includes a high-resolution nature scene with a collection of other smaller video clips embedded into the backdrop to demonstrate the company’s streaming breakthrough. The company streamed the video over a standard home Wi-Fi network to a Samsung Galaxy S7 inside a Gear VR headset. The S7’s display is capable of producing dual 1280 x 1440 images with 90-degrees FOV, which roughly translates to a maximum resolution of 6K for a 360-degree video.

Visbit’s VVSO technology takes advantage of the full resolution of the clip by offering zoom functionality. If you double-tap on the trackpad on the Gear VR, the image zooms in to show the full details of the 12K image. With this function, even lower resolution displays can benefit from the video that exceeds the resolution of the panel.

Visbit’s streaming technology currently supports Samsung Gear VR and Google Cardboard. The technology isn’t readily available yet, but the company is running a closed beta with select creators and plans to launch an open beta later this year. See the Visbit website for more information about View-Optimized Streaming and how you can sign up for the beta.

As frequent Tom’s Hardware readers know, even though we’ve enjoyed copious VR demos of late and are free to describe them in loving detail, we’re rarely allowed to show any footage or images of the experiences. That changed at Mobile World Congress, where we were allowed to shoot the IEEE’s Mars Rover demo.

IEEE?

Yes, that IEEE. Why is the IEEE doing VR? There are two reasons: Primarily, the group believes that shared VR experiences will be a key element of widespread VR adoption. It’s hard to disagree with that assessment — and really, if you think about the concept of virtualized spaces, engaging with others within them is a fundamental use case. In addition to the fact that it enables more powerful interactions for people working on projects together remotely, it sounds thoroughly enjoyable, doesn’t it? There are other companies working in this space, including AltspaceVR.

As to why the IEEE chose to show what it can do in multiplayer VR with a Mars rover demo, it’s simply because many of the engineers engaged with it formerly worked on projects for real space missions. Therefore, the Mars Rover experience is a love letter, if you will.

Mission On Mars

Two of us got the chance to navigate through the Mars Rover demo. In the video, you’ll see our Editor-In-Chief, Fritz Nelson, but I had a go at it as well.

There are three participants at a time. Each of you is seated and dons an Oculus Rift DK2 headset and some headphones. You each get a right-hand joystick and left-hand thruster. That sounds terribly exciting, but you’re limited to moving in one of four directions on the joystick and forward and backward with the thruster. (In other words, there’s no Martian sniping to be had.)

A captain (a fourth party who is not in the virtual environment with you) doles out instructions. Each player has four tasks to perform, so there are 12 total. When the demo begins, you’ll get a notification on your HUD that a given task must be performed, and only you can see it. Usually that task is one that you aren’t able to perform yourself, so you have to tell a teammate to do it. That requires you to shout the task out so that a teammate will hear you and perform it; meanwhile, your teammates will be shouting out instructions that you (and your other teammates) need to perform. If the tasks are not performed in a timely manner, the mission will fail and everyone dies.

In my run, none of us actually spoke the same native language. Correctly assessing my monolingual gringo-ness, the captain asked, “Alejandro, do you speak English?” He did. I apologized for my terrible Spanish, in terrible Spanish. “Mi español no es bueno,” I admitted sheepishly. Our other compadre spoke enough English for all of us to get by.

And get by we did. The Mars Rover experience is an on-rails experience, so it rolls along no matter what you do. It was surprisingly difficult to read each task that popped up, process if it was one of mine or one of my teammates, and either fix it or shout it out to the crew, all while listening for instructions from my teammates on my objectives. And, of course, I had to perform my tasks, too.

Our stellar crew was able to solve all of the little problems as they arose, and we made it out alive and with our vessel intact. Our captain congratulated us on a job well done.

Portending

The Mars Rover demo was simple, and although it was fun, it’s not the kind of thing I wanted to do again and again. But for the IEEE, it seems that this is really just an engaging proof of concept. Is multiplayer VR feasible? Yes. Can it be fun? Sure. Could it work locally as well as remotely? Indeed.

As I mentioned earlier, the IEEE isn’t the only group working on multiplayer VR — that is, fluid, active, unpredictable experiences, not just games where you shoot things together — but it does see a (near) future where people can collaborate on projects, communicate with each other, and otherwise engage within virtual environments. 

Update, 3/8/16, 1:24pm CT: Globalcore worked on developing this experience.

Seth Colaner is the News Director for Tom's Hardware. Follow him on Twitter @SethColaner. Follow us on Facebook, Google+, RSS, Twitter and YouTube.

Introduction

Networking exists in almost every household, enabled through a mixture of wired and wireless adapters. Wired Ethernet adapters, specifically RJ45-based, have been available for years and on many different devices. But are all of them the same? If so, does anything really set one apart from another? After all, it seems like almost every motherboard comes with some form of network interface adapter. Perhaps for this reason, network interface cards (NICs) are some of the most overlooked, yet essential components in terms of expanding a computer’s functionality.

The process begins with determining what your data requirements are so that you can figure out which capabilities are important. If you are a gamer, you might want to prioritize low latency. Conversely, if you are more worried about streaming media, then quality of service will be a higher priority. Purchasing the wrong network interface card could cause your data to bottleneck. Fortunately, with a basic comprehension of the choices available to you, this can be prevented.

Popular Network Interface Types

Network interface cards, motherboard-down and USB network adapters are the most prevalent wired interfaces on the market today. It is important to recognize each device's strengths and weaknesses to understand how NICs are different from other available solutions.

The NIC is a slot-based card that connects directly to the motherboard of the computer. The major benefit to these cards is that they handle the buffer storage, encoding and decoding of data through the seven network layers. This allows the processor to concentrate on other tasks, rather than the data being received by it. Typically, there are more advanced feature sets available on network interface cards, as they have dedicated storage and the processing to support this without affecting the CPU. Conversely, an installed NIC will take one of your motherboard expansion slots, which can be troublesome if you are running several add-on components in a small form factor computer.

Motherboard-down controllers allow you to provide networking services without utilizing an expansion slot. They're typically less expensive than an add-in card, though.

The downside is that they do utilize processor power to encode and decode data through the network layers. When dealing with a high amount of traffic, especially encrypted data, this could become noticeable. It should be noted that most motherboard manufacturers do direct this traffic through the southbridge.

Wired USB adapters provide portable and quick installation for network connections that can be installed by individuals of any skill level. They also facilitate the ability to add a connection without requiring a motherboard change or filling one of its slots. Conversely, USB is traditionally slower than the quickest Ethernet links, causing potential bottlenecks when network performance exceeds 50 Mb/s using USB 2.0. USB 3.0 should alleviate that.

Hardware Differences

The most current type of media in household networks is Cat 6 Ethernet cable connected to devices using RJ45 connectors. Also known as 8P8C (eight position, eight contact), the connector ends of the Ethernet cable look like a phone jack connector, only larger and with eight wires instead of four. This is important to remember when making a selection because fiber optic network interface cards are available at a much more reasonable price. While the fiber NIC devices have come down in price, typically routers and other networking components that are fiber-based are still on the higher side of the price spectrum.

In terms of the RJ45-equipped devices, most only come with one available port. But there are those that can host up to four ports on one NIC. These are typically for business environments, but can be extremely helpful for busy and very powerful home media servers using a technology called network bonding. Essentially, this allows data to be transmitted over more than one cable to allow for a higher throughput and less latency for file access.  If you're connected to multiple network devices, they can also provide fault tolerance in the event of device failure. Typically, unless the server is being accessed by more than three devices at a time on a regular basis, network bonding should not be required.

Two major interfaces are available to desktop PCs, depending upon the motherboard’s interface. Unfortunately, they have similar names, so the nomenclature can be confusing if you are not familiar with them. The older standard, PCI, carries a double-sided 62-pin zero insertion force (ZIF) socket. This standard is still being produced and developed on most motherboards, but not all. The newer standard is PCI Express, shortened to PCIe. You'll find PCIe slots in x1, x4, x8 and x16 configurations. Most add-in NICs drops into a single-lane slot with 18 pins. It is important not to confuse PCIe x16 and PCI slots; PCIe x16 is slightly longer and often has a card snap towards the end of the slot.

Software Differences

Each manufacturer develops its own firmware to execute instructions. These differences are what give each vendor's products unique characteristics. Certain companies employ proprietary instruction sets, which are devoted to ensuring minimal latency. Other manufacturers focus on security and features like better error correction. While the purpose behind firmware is to instruct the hardware on how to perform its actions, it is as vital as the physical components on the board.

When I am looking for a new card, I often look to see which companies are providing updates to their firmware at a regular interval. This leads me to believe they are striving to maximize performance beyond what was first launched. Firmware releases are often found on the support pages of each NIC manufacturer’s website.

It is important to perform research on a specific manufacturer’s features to determine if they only work with other devices made by the same vendor. Since there is a standardization of data transmission, most efficiency increases fall outside of this standardization, making them proprietary. Therefore, spending extra money on a desired feature may be futile if purchased without the coupling router and matching network cards to other machines.

Anatomy

The main function of a network interface card is to encapsulate data and translate it into electronic signals to be sent on the Ethernet cable to its destination. In order to complete this task, the data originates from the motherboard, which transfers information to the device through its expansion slot. As discussed, the current standard is PCIe, and most NICs drop into x1 or x4 slots. It is still possible to purchase PCI-based NICs for backwards compatibility. Standard PCI slots can still be found on modern motherboards, allowing legacy cards to work in newly upgraded computers. 

Once the data is on the card, it is directed towards the controller chip, which is similar in concept to a low-end CPU. Data received by the controller is read and encapsulated with the information of the destination prior to being sent out on the line. If the data is received from the Ethernet cable, the controller is responsible for stripping the first three layers of encapsulation prior to handing the data off to the computer. The RJ45 jack is responsible for generating and receiving the electronic signals that are put on the network cable. 

The firmware is stored on a chip known as an EPROM (erasable/programmable read-only memory) that can be manually updated to provide more functionality and code fixes. 

Practical Applications

While wireless is a major facet of an average home network, wired network devices still have extremely practical uses. The largest benefit of wired networks is the consistent amount of bandwidth they provide, lower latency and minimal environmental concerns. One of the major limiting attributes is, of course, a lack of mobility. As a result, it is important that factors that would make a wired network solution more beneficial than a wireless solution are understood.

Environmental factors can cause severe performance issues with wireless networks depending on building material, distance from the router and other devices saturating the wireless bands. Thicker materials like brick and steel-framed construction often found in commercial and dormitory environments diminish the performance of wireless. In addition, larger areas being covered can require wireless repeaters to gain useful signal strength throughout larger homes or those with multiple floors. This can cause data loss and poor latency when re-transmitting data from one end of the wireless range to the other.

Depending on the wireless network spectrum you utilize, the number of consumer products that operate on the same band can cause poor performance. Wireless telephones operate on the same 5GHz and 2.4GHz range as wireless routers. This can be combined with all of the routers and cordless devices from not only your house, but those of your neighbors as well. Since 2.4GHz only operates off 11 channels (or bands), and 5GHz has 22 channels, it is easy to see how wireless data collision could occur in densely populated buildings.

In these situations, it is preferable to operate through a wired network connection to eliminate network degradation. One major facet of planning your wiring placement includes avoiding electrical wiring as much as possible, since electromagnetic fields can cause issues with data delivery. As well, the length of the run is important, as 100 meters is the standard when running Category 5, 5e and 6 twisted-pair cabling.

High bandwidth availability provides file sharing and media servers with a significant performance enhancement over those utilizing wireless technology. Even providing a direct router connection utilizing a wired network can give wireless devices higher-quality file transfers and streaming. This is because the media is only streaming through the wireless channels once rather than from the media/file server to the router, then to the client requesting the file or media. Ideally, you would want to use a wired network connection between all points from where the media or file is to the computer requesting that data.

To gamers, a wired network connection to the modem provides much lower latency, improving the gaming experience. A ping response is essentially the amount of time it takes for a data packet to get sent to a networked destination like a router or server and then return to your computer. As previously mentioned, wireless connections can be affected by many environmental variables that can cause your ping to take longer than a wired connection. Other factors, such as available Internet bandwidth, do influence this, but a wired network connection is your best bet for minimizing delays.

Speeds

The ability of a network to perform well demands on the devices attached to the infrastructure. Standard home network cards typically are available with two data rates: 100 megabits per second (Mb/s) or 1000. It is important to note that gigabit is another term that is often interchanged with 1000 Mb/s speeds and they are both used frequently. Essentially, this is how much theoretical data could be transferred through each link.

Another concept that is important to recognize, but not as prevalent as it was in the past, is duplex. Duplex essentially is the ability for data to pass simultaneously in both directions at the same time. The term full duplex is used when data travels down both the send and receive portions of the cabling. Half-duplex means that data may only travel down the wire in one direction at a time. Essentially, running at half-duplex cuts potential bandwidth over 50 percent, since it is consistently waiting for the line to open to transmit the next piece of information. While it is rare to find a card that only support half-duplex, it is important to ensure any new device does offer full-duplex functionality.

Settings

While each system and network performs differently, a few settings could assist with maximizing your throughput. For example, disabling network energy efficiency in the detailed power settings and disallowing the device to go to sleep through the advanced device settings does impose higher power consumption, but those toggles also allow that device to perform at its full capability at all times. If you are operating on a network where wait times seem relatively long, then auto negotiation may be an issue. Change your speed and duplex to match the router and network interface card's capabilities. Personally, I utilize the 100 Mb/s full-duplex setting on both devices.

Common Chipsets

The network card's controller is the piece of hardware that encodes and decodes the frames of data sent through the wire. In addition, while each layer of the network model is theoretically the same, each manufacturer has a different set of instructions that operate at varying efficiency levels. Therefore, buying a NIC powered by an ubiquitous controller vendor can improve your networking experience.

While one of the most popular names is recognizable for manufacturing other components, the remainder tend to specialize in networking. Of course, Intel is that household brand, which provides a solid chipset with interesting features. Other major players in the NIC chipset market are Broadcom, RealTek, Marvell Technology and Qlogic.

Often, the manufacturer of the NIC itself is not the chipset manufacturer. As with AMD or Nvidia and its board partners, the controller is often purchased and placed on another company's PCB. Another important factor is that if a manufacturer develops a proprietary technology, the feature typically only works with other compatible devices. Therefore, it is important to perform research while determining the network interface card you are looking to purchase.

The Future

We are not far off from having the ability to transfer one terabit per second across the network line. Within the past five years, the IEEE has approved a wired standard for 100 gigabits per second data transmission, and continues to progress on the standard, updating it as recently as 2014. The current speeds available for wired network interface card technology allow for 10 Gb/s with a short run of Cat 6 cabling. It is fully expected that 100 gigabits per second will be a reality utilizing standard RJ45-connected NICs.

While this standard is not ready for consumers at theorized speeds through copper wire, manufacturing processes in fiber optics have made that technology more affordable for the average consumer. This could potentially be a reason for the delay in development of the copper equivalent, since fiber optics can currently produce speeds beyond the current 10 Gb/s of available RJ45-attached NICs. Unfortunately, fiber network interface cards still start around nine times the average cost of RJ45 network adapters. With the current trend in networking development, it is not a question of whether fiber will become the main household standard of network interface cards, but rather when this will happen.

Conclusion

Turning back to the original question, where we asked if all NICs are the same, many of their characteristics are in place due to the nature of IEEE standardization. From the standpoint of functionality, they are similar. But depending on your existing hardware, the device that you require can be slightly different. Yet it is possible that these differences can make or break the maximum efficiencies of your network.

Remember to start by researching your existing network and developing an understanding of its capability. This will allow you to determine the correct network interface card to purchase for your machine. As well, remember to keep an eye on the vendor-specific features, as previously mentioned. Ensuring that you purchase the correct device for your computer and those around it allows you to provide the best possible experience from your network.

MORE: Networking in the Forums

Brent Woolverton is an Associate Contributing Writer for Tom's Hardware.

Follow us on Twitter @tomshardware, on Facebook and on Google+.

In the August 2012 IEEE 802.3 Ethernet Bandwidth Assessment report, the Institute of Electrical and Electronics Engineers (IEEE) predicted that networks will need to support an average of 58 percent compound annual growth rates (CAGRs) with capacity requirements of 1 Tb/s in 2014 and 10 Tb/s in 2020.

To meet the requirements of ever-increasing number of users, Internet enabled devices and services such as video on demand and social media, the IEEE has launched the IEEE 802.3 "Standard for Ethernet" study group to explore the development and 400 Gb/s Ethernet standard to effectively manage this exponential growth in bandwidth requirements.

"Traffic is growing everywhere—more Internet users, more ways to access the Internet more quickly, higher-bandwidth content, new applications enabled, etc.—and it's critical that we move now to create a plan for the Ethernet ecosystem to evolve beyond today's capabilities, in order to accommodate the burgeoning bandwidth tsunami," said John D'Ambrosia, chair of the new IEEE 802.3 400 Gb/s Ethernet Study Group and chief Ethernet evangelist, CTO office, Dell. "The launch of this study group is the next critical step in evolving the IEEE 802.3 standard to stay ahead of industry's needs. It builds on two years of open efforts around inviting Ethernet's vast array of stakeholders into the work of assessing and tackling the market's emerging application requirements."

The IEEE 802.3 "Standards for Ethernet" Study Group will be meeting at the Fairmont Express in Victoria, BC between May 14 and May 17, 2013. Those interested in taking part can view further information and register for the event at its event page.

_{Contact Us for News Tips, Corrections and Feedback}

IEEE 802.11aa-2012 includes a more sophisticated prioritization to balance network management information and data traffic to achieve better network performance as a result.

Also, 802.11aa integrates new Quality of Service (QoS) features, including group-addressed transmission service, stream classification service, management of overlapping networks and support for the IEEE 802.1Q Stream Reservation Protocol. According to the IEEE, the changes deliver improved management of audio/video stream quality, link reliability and application performance when channel capacity is insufficient or when WLANs overlap on the same radio channel.

“IEEE 802.11 is constantly being refined and innovated to address the emerging needs of the marketplace, and these two new standards are further proof of that fact,” said Bruce Kraemer, chair of the IEEE 802.11 working group. “Both of these amendments specify timely enhancements to the base standard that continue to expand IEEE 802.11’s use in more and more business and residential applications around the globe.”

The IEEE 802.11aa is available for purchase at the IEEE Standards Store. The PDF is priced at $5, while the printed version is sold for $99.

_{Contact Us for News Tips, Corrections and Feedback}

Can LGA 2011 Be Made More Affordable?

Motherboard designers can approach cost-cutting from several different angles. They can selectively lop off certain features or change their layouts, facilitating fewer PCB layers. They can pare back value-added bundles or even offer mail-in rebates that a majority of folks never take the time to redeem. We've seen these approaches work really well on mainstream platforms, and we're sometimes surprised to find exceptional reliability persisting, despite the need to shave off precious dollars and cents. But Intel's LGA 2011 interface is a different animal entirely. Its four memory channels feed a complex processor, which attaches to a more expensive PCH. Together, that all results in a higher starting price for Core i7-3000-compatible motherboards.

And then you have to take the cost of Intel's processors into account. Even the lowest-end Core i7-3820 is expected to sell for close to $300, and that's a quad-core mode. Consequently, we can't expect too many enthusiasts to tolerate a cut-down, incomplete motherboard to complement any Sandy Bridge-E-based chip. 

Pricing and quality expectations come to a head in the sub-$260 LGA 2011 market, where some buyers will certainly wonder why an entry-level board still costs $100 more than LGA 1155-based platforms with seemingly similar on-board extras. Quite simply, the costs associated with Sandy Bridge-E are higher, in part because of Intel's prices and also because the boards are more difficult to design.

.

Feature tables tell a lot about what each motherboard offers. But design and implementation make some features more valuable than others. We’re now ready to examine and detail the design, performance, and overclocking capabilities of these six not-quite-economical products.

Before we move on to the next page, though, we want to point out that, finally, all six of these vendors are on-board with similar three-year warranties, satisfying our former criticisms about a lack of value in the form of support.

ASRock X79 Extreme4

ASRock’s continued effort to provide higher-end hardware without letting this platform's inherent price premium get out of control is easily illustrated in an X79 Extreme4 that has three extra SATA 6Gb/s ports, eSATA, dual-controller USB 3.0 for internal and I/O ports, and a full set of O/C-friendly features.

Those O/C-friendly features include a Port 80 diagnostics display, an I/O panel CLR_CMOS button, internal power and reset buttons for bench testing, and a eight-phase CPU voltage regulator with oversized cooling.

Front-panel USB 3.0 and one of the added SATA ports are found above-center on the motherboard’s front edge for easy reach of front-panel connector cables. Conversely, the front-panel audio cable is located as far away from any front-panel connectors as it could possibly be, at the rear of the X79 Extreme4’s bottom edge. Remaining ports are designed to coexist with up to three double-slot graphics cards, and ASRock even adds an extra space between the board’s two true x16 PCIe slots for additional graphics cooling.

The provision of a full feature set on a low-cost board is never free of compromise, as the X79 Extreme4 has only four memory slots. Although you'd have to decide right away whether to use 2 GB, 4 GB, or 8 GB modules to achieve quad-channel functionality without any option for future memory expansion, low memory prices make this a reasonable production-cost concession in our minds.

This is the first ASRock Extreme-series board in recent memory not to include the firm’s USB 3.0 front-panel bay adapter. We partly credit case manufacturers with making this part unnecessary, as removing it allows ASRock to further reduce cost and pricing.

We would have liked to see more than four SATA cables, however, even though this is probably enough to serve the budget-enthusiast market.

X79 Extreme4 Firmware

ASRock attacks the overclocking market on two fronts, with “CPU EZ OC” profiles up to 5.0 GHz, plus full manual controls. We recommend keeping any of Intel’s 32 nm CPUs below 1.40 V for increased longevity, and we perform our own comparative analysis at 1.35 V.

After suffering through two motherboard round-ups with a CPU bug for which only one of ASRock’s competitors had a workaround, we finally moved on to a new bug-free processor based on Intel's newer C2 stepping. Choosing 47 x 100 MHz gets us to a respectable 4.7 GHz, though the board’s 1 MHz BCLK increments prevented us from seeking a few MHz more (we would have liked to see more granularity there).

Memory data rates up to DDR3-2400 are available at stock base clock, though most Sandy Bridge-E processors require advanced tweaking to reach this data rate.

Voltage controls include CPU core and CSA, the latter assisting primarily in memory controller overclocking. Benchmarked at default settings, our memory gets bumped up to its recommended 1.65 V when overclocked.

Unlike most entry-level boards, the X79 Extreme4 adds secondary and tertiary timings to its manual memory configuration.

Asus P9X79

Asus puts all four of the P9X79’s USB 3.0 ports on the rear panel, eliminating the possibility of a front-panel header. The board does, however, retain eSATA and IEEE-1394 ports from its more-expensive siblings, along with Asus’ USB BIOS Flashback and Quick BIOS Screen Capture features.

USB BIOS Flashback uses a custom IC to enable firmware flashing from a USB drive with no other hardware installed. That’s particularly beneficial to users who actually need a firmware update to simply support hardware they bought with the board. As you'll see shortly, this is a very real concern, and a competitor was affected by its inability to accommodate our C2-stepping processor right out of the box.

The P9X79 beats several competing products in maximum memory support by including eight slots. While the board also gives up any additional SATA ports, most enthusiasts will expand beyond four modules long before they see the need to expand beyond the X79 Express' six SATA ports.

The P9X79 includes switches and a button for Asus’ EPU, TPU, and MemOK modes. EPU is an automatic underclocking routine for enhanced power savings under low loads, TPU is an automatic overclocking routine, and MemOK temporarily underclocks memory to ease booting with improperly-programmed modules.

Like most other entry-level X79-based platforms, the P9X79 relies on Sandy Bridge-E’s 40 PCIe 3.0 lanes to support up to three graphics cards with x16-x16-x8 connections. Asus pushes the x8 slot to the bottom of the board, however, where installing a dual-slot graphics card requires an eight-slot case (in addition to smashing flat any cables connected to headers along the board’s bottom edge). This is particularly frustrating for users who hoped to take advantage of Asus’ Q-Connector cable bundling feature. Dual-card users fare better, since the board provides an extra space between its two x16 slots for improved airflow.

Though the P9X79 supports three graphics cards, it does not include a three-way SLI bridge. A two-way SLI bridge, I/O shield, and four SATA cables complete its installation kit.

P9X79 Firmware

P9X79 firmware includes several automatic and manual overclocking options in addition to Asus OC Tuner, which is designed to automatically find the company's idea of an optimized overclock based on incremental clock increases and stability tests. We’re a little cautious with core voltage settings, though, and prefer to set this manually.

We didn’t quite reach 4.7 GHz on the P9X79, but small 0.1 MHz BCLK increments allowed us to set 4672 MHz using the same 47x CPU multiplier.

The P9X79 plays on our love of simplified menus by placing key voltage controls on the main Ai Tweaker menu. Per-channel DRAM reference voltage levels might be perceived as over-the-top, however, since only the most persistent tweakers use these.

Choosing “Ultra High” CPU Load-line Calibration from the DIGI+ Power Control submenu allows us to retain a CPU core voltage very close to our targeted 1.35 V under heavy processor loads.

Disabling EIST within the CPU Performance Settings submenu let us lock in a constant 47x CPU multiplier, exceeding non-Turbo Boost ratios even though the menu shows Turbo Mode disabled.

The P9X79’s DRAM Timing Control submenu adds slew rate to the familiar primary, secondary, and tertiary timings.

Biostar TPower X79

Often perceived as a contender for extracting some of the best overclocking value, Biostar stepped its game up a bit with its TPower X79 by providing twice as many USB 3.0 ports compared to most competitors. We find six of those connectors on the I/O panel and two available via front-panel header.

Biostar's customers aren’t even forced to give up a bunch of other features to get those extra ports, since the board still features a third-party SATA/eSATA controller, three PCIe 3.0 x16 slots electrically wired to run at x16-x16-x8, a Port 80 diagnostics display for overclocking failure analysis, and integrated power/reset/CLR_CMOS buttons for bench testing.

The TPower X79 even has two high-capacity EPS12V connectors to feed its mid-capacity voltage regulator. Perhaps the most significant compromise is a quartet of memory slots, giving you one shot to grab the quad-channel kit you need and no room for an upgrade. With a Web price of only $230, that sacrifice is fairly small in exchange for such a wide range of added controllers.

Unfortunately, installing a third graphics card will be problematic for performance enthusiasts, since the extra-stiff USB 3.0 front-panel connectors of most cases block the installation of most cards. As with Asus' board, two-way CrossFire and SLI are preferred, and Biostar provides an extra space between the top two x16 slots to assist airflow between a pair of boards.

Competitors could learn a thing from Biostar about low-cost installation kits, as the TPower X79 includes six SATA cables to support all of the chipset’s native ports. But perhaps Biostar could also learn something from ASRock, the company that enables three-way SLI with a bundled bridge and smarter header placement. The TPower X79 is almost exclusively focused on dual-card graphics configurations, though going the single-GPU route is certainly an option too.

TPower X79 Firmware

Biostar copies over its familiar O.N.E. menu from earlier BIOS implementations to its UEFI with a few modifications, such as moving the menu bar to the bottom of the GUI.

Unfortunately, this particular board was uncharacteristically difficult to overclock compared to the firm’s other recent efforts, with nonfunctional boot failure recovery and a lack of overclocking profile storage that forced us to continuously clear all settings and then reset everything to known-good values in our overclocking attempts. We finally reached a stable (but frankly pretty weak) 4455 MHz overclock using a below-stock 99 MHz base clock with a 45x multiplier.

We also had to set a 1.37 V Vcore to achieve something close to 1.35 V.

Primary and secondary memory timings are also available from the main O.N.E. menu.

Though voltage settings are few, core, DRAM, and CSA voltage are all available. Unlike the CPU voltage found further up its menu, this CPU core voltage setting is an offset that complements the baseline figure.

ECS X79R-AX Black Deluxe

ECS' X79-based branding is a little strange. Its “Black Extreme” moniker applies to a high-end part labeled X79R-AX, while its “Black Deluxe” moniker goes to the mid-range part officially called the X79R-AX Deluxe. Those official names come from both the company's Web site and the descriptor labels applied to its boxes. In other words, the Deluxe is a lower-end model than the non-Deluxe, and the word Extreme rarely appears on the top-end variant. In order to keep things simple, we’re ignoring the word Black and going by the label’s official X79R-AX Deluxe designator.

Deluxe could refer to the board’s support for four graphics cards, its four rear and two front-panel USB 3.0 ports, or perhaps its quad SAS 6Gb/s connectors. Two of those features are unique amongst today's contenders.

Pathway switches allow the X79R-AX Deluxe to go from two full-bandwidth PCIe 3.0 x16 links to four x8 connectors by sending eight of each grey slot’s lanes to the corresponding white slot. That only happens when the switches detect a card in the white slot, though.

Of course, four-card arrays aren't for everyone, and they may not even be for anyone shopping in the X79R-AX Deluxe's price range. They certainly can't be used if you also want front-panel USB 3.0 support, since the front-panel header is placed too closely to the fourth graphics slot for a card and a cable to coexist. Three-way CrossFire is still a solid possibility, but the slots are most suitably situated to support two really big graphics coolers.

The CPU’s remaining eight lanes of third-gen PCIe are reserved for on-board devices, with four of them providing a communications pathway for the PCH's SAS controller. These are the same ports Intel decided to disable on desktops, and ECS was the only manufacturer to enable them anyway. These must be enabled in both firmware and by Intel drivers supplied by ECS. Mixed tales of triumph and tragedy have been attributed to the use of these ports, and the tragedies are mostly likely related to the fact that Intel decided not to support them.

In the days to come, Intel will introduce the server/workstation version of this same PCH, which will be largely identical to the desktop version, aside from its name. That'll be the product to look to for official Intel SAS support.

The X79R-AX Deluxe officially supports six drives and actually includes six SATA cables. ECS adds a single SLI bridge. You'll need to look to your graphics vendor for a CrossFire bridge.

X79R-AX Deluxe Firmware

As ECS continues developing its firmware, we’re looking at a product that appears nearly perfect and full of value. Of course, appearances can be somewhat deceptive.

The M.I.B. X main menu provides all of the chipest ratio and voltage controls we require, disappointing us only in that those voltage settings are represented by added or subtracted voltage, rather than fixed-mode values.

The CPU Overclocking submenu includes base and Turbo Boost ratios, along with voltage increases for Turbo Boost and current limit overrides.

Memory ratio, primary, and secondary timings are also available from the Memory Overclocking submenu.

All of those options should be enough to get us to the limits of our CPU, or at least this board. However, the shipping firmware’s inability to support our C2-stepping processor put a wrench in our overclocking plans. ECS did hurry to send us a replacement firmware, which did the trick, but presented us with a couple of other issues.

Yes, the replacement UEFI worked. But giving ECS another shot after our cut-off date for submissions would have forced us to re-do all testing on the other five boards as well. Plus, there's the fact that end-users don't have the luxury of swapping out a new C2-based processor for a C1 chip to flash a new firmware. This is a compatibility issue that'll create sticky situations until all of the X79R-AX boards available include ECS' latest build. The consolation that we offered was a chance to participate in our benchmark suite using the older C1-stepping CPU. We simply couldn't count our overclocking attempts on two different processors, though.

Intel DX79TO

Intel now counts itself amongst the motherboard vendors with thorough overclocking functionality. Its DX79TO stands out most clearly for its lack of passive voltage regulator cooling compared to the other boards tested in this round-up. A fairly large heat sink positioned where you would have expected to find a northbridge is nothing more than an extension of the PCH cooler that alleviates any need for a chipset cooling fan.

Intel adds two USB 3.0 ports to its rear panel and not much else. Though a Port 80 diagnostics display, a power button, and a reset switch all aid in bench testing, IEEE-1394a is its only other concession to add-in connectivity.

We would expect Intel’s reduced-price motherboards to maximize the platform's built-in capabilities, and were pleased to find eight memory slots available for quad-channel upgrades. On the other hand, a mere three 1/8" analog audio jacks on a board that lacks even a single digital output is far more disappointing, and we're left to wonder where the CPU's eight remaining lanes--normally reserved for a third graphics slot--have gone.

At least the board still includes Intel’s Back-to-BIOS button, which allows the system to boot at its default configuration, while still retaining custom settings in its firmware GUI.

While a mere two graphics slots is likely sufficient for most enthusiasts, those same enthusiasts could save even more money by simply switching to LGA 1155. Slot connectors are relatively cheap, and the eight lanes that most competing products use for a third slot are native to all LGA 2011 CPUs. Intel doesn't even take advantage of the reduced graphics support by spacing remaining slots farther apart for increased graphics cooling. Though we're certain some LGA 2011 buyers are simply looking for a low-cost yet reliable way to support Intel's latest six-core processors, competing platforms that include both an extra graphics slot and extra space between the two primarily slots should be far more attractive to the majority of buyers.

A board with fewer internal headers should inherently be easier to lay out, though we would have preferred to find the front-panel audio header a little further forward from its bottom-rear corner placement. The cables of some cases come up less than an inch short of reaching this location.

Four SATA cables and a single SLI bridge illustrate Intel’s focus on cutting cost in the DX79TO’s least-significant places. These savings add up to a board that comes in as the least expensive of today’s competitors, in spite of Intel’s reputation for arming its boards with stability through thorough validation.

DX79TO Firmware

Intel’s atypical GUI is easier to use than a number of competing firmware implementations, though it does take a few minutes to master. The main “Performance” menu, for example, shows many settings, but only a few are selectable.

Host Clock Multiplier refers to the setting formerly known as a boot strap on older LGA 775 boards. This sets the ratio between the X79 PCH and CPU base clock. Other settings include base clock, CSA voltage, I/O and PLL voltage, and PCH core voltage.

CPU core voltage is found in the DX79TO’s Processor Overrides submenu, along with V-Droop control, current limits, and Turbo Boost ratios. Setting Runtime Turbo Ratio to Enable allows a single Turbo Boost multiplier to be adjusted from within Windows while reducing firmware GUI ratios to a single setting. We used the 47x ratio with a slightly underclocked 98.88 MHz BCLK to reach a stable 4.616 GHz overclock.

The DX79TO’s Memory Overrides submenu includes memory voltage and multiplier, plus primary and secondary timing controls.

MSI X79A-GD45 (8D)

MSI rivals Asus in offering the fewest extras for the money by integrating two USB 3.0 controllers and chipset-only SATA (compared to Asus’ eSATA ports and lack of front-panel USB 3.0). Both boards support triple-card CrossFireX, though, and both have a full set of eight DIMM slots.

It bears mention that the X79A-GD45 (8D)’s extra USB 3.0 connectivity is a little less valuable than Asus' eSATA because you can't use the front-panel USB 3.0 header with three graphics cards installed. Short add-in cards fit. Thin add-in cards fit. But USB 3.0 front-panel cable ends are simply too stiff to fit under a graphics card heat sink.

The other two x16-length slots each have a single PCIe 2.0 pathway suitable only for low-bandwidth devices. That functionality could include low-end graphics cards serving 2D needs.

Apart from the unfortunately-positioned USB 3.0 front-panel header, we have no major objections to the X79A-GD45 (8D)’s layout. It's somewhat inconvenient to have the front-panel audio header positioned a little too far back for some of our cases. That's offset by a minor design triumph in MSI's inclusion of a manual switch for its dual-BIOS function.

Supporting up to six internal SATA drives, the X79A-GD45 (8D) includes all six SATA cables. MSI’s M-Connector cable bundler, a single SLI bridge, and an I/O shield complete the installation kit.

X79A-GD45 (8D) Firmware

Overclockers generally know what settings they want to change. Meanwhile, MSI sets its firmware's fonts really large for the most trivial settings and really small for the most important settings. Could that reverse emphasis be designed to discourage more casual users from playing with the wrong option?

All primary clock and voltage controls are found within the main OC menu, including base clock, chipset ratio (Clock Strap), CPU multiplier, and DRAM ratio.

CPU Core, CSA, VTT, and PLL voltages are all adjustable.

If you choose not to manually specify CPU voltage but still want a little more performance using Intel’s stock voltage scheme, use the separate Override Voltage option. DIMM voltage is adjustable in channel pairs, while DIMM reference voltage is further divided into single-channel settings.

While Turbo Boost ratios are adjustable based on the number of active cores within the CPU features menu, our target was to simply find one stability limit with all cores loaded. Hitting that target was as simple as setting the base ratio to 47x in the main menu, pushing our CPU to 4.70 GHz.

Primary and secondary memory timings are adjustable from the Advanced DRAM Configuration submenu. Shown timings correspond to XMP values for our test modules.

Test Settings And Benchmarks

A credit to Swiftech’s reliability, its LGA 775-era Apogee GTX kit keeps our CPU temperature well below its thermal throttling limits under every test condition, even while using its original coolant and nothing more than a bracket and screw change to fit the newer LGA 2011 platform. We added an SFF21D cooling fan above its water block to assist voltage regulator and DRAM cooling.

G.Skills Ripjaws DDR3-2200 allows us to test the overclocking capabilities of each motherboard at one module per channel, and the basic performance of each board at default DDR3-1600 SPD values. The firm sent a second set for our eight-DIMM tests using the same ICs, but different XMP values.

Nvidia’s GeForce GTX 580 pushes our CPU limits under games, enhancing any small performance deficits that some motherboards occasionally exhibit.

Samsung’s MZPA256HMDR 256 GB SSD keeps load times and idle power low.

Benchmark Results: DiRT 3

For most games, our lowest settings show differences in CPU and/or memory performance, while our highest settings show differences in GPU and/or PCIe performance. We expect substantially similar results across all tests, but look for any large gains to expose less-than-honest default clock rate settings or large losses to expose implementation problems.

ECS leads DiRT 3, followed closely by Intel, but no evidence to suspect anything out of the ordinary. Both boards show spot-on base clock frequencies in CPU-Z.

Benchmark Results: Metro 2033

Metro 2033 chokes most systems, making this already-old game an excellent replacement for the ancient, original Crysis. Once again, we look for large differences to expose non-standard clocks or configuration issues.

The leaders from DiRT 3, ECS and Intel each take top positions in one of the two test charts. Typical variations of roughly 0.5 FPS between consecutive runs point to performance that’s very close to par for all tested boards.

Benchmark Results: StarCraft II

ASRock and Biostar lead in StarCraft II, though those leads are within the typical performance variation between consecutive runs (around 2-3 FPS). Those leads vanish completely at 2560x1600.

Benchmark Results: Audio And Video Encoding

Though we haven’t seen any consistent performance leadership up until this point, both of our audio encoding applications favor the same three motherboards. Both applications are also single-threaded, scaling based on the clock rate of a single core. We didn't see a noticeable difference in base clock, though, which would have allowed one platform to run slightly faster than the others.

HandBrake shows no performance difference between motherboards.

One of the three motherboards to top our audio transcoding charts, Asus’ P9X79, leads again in MainConcept.

Benchmark Results: Productivity

Performance differences in Adobe Photoshop are too small to detect during a one-minute project.

Biostar picks up a miniscule lead in 3ds Max.

Two of the three leaders from our audio encoding benchmarks retake top positions in file compression.

Biostar’s TPower X79 picks up its second productivity win in ABBYY FineReader.

Power, Heat, And Efficiency

Intel demonstrates the lowest idle power consumption, while MSI matches it, on average, with the lowest full-load power consumption. ECS' idle power result could be slightly off due to the fact that we had to use a C1-stepping processor with its qualifying firmware revision.

A big heat sink on MSI’s voltage regulator, along with moderate power consumption, contributes to impressive thermals at stock CPU settings. Biostar’s second-place temperature is equally notable given its higher power consumption.

With less than 1% performance difference separating various motherboards, efficiency is primarily a reflection of power consumption. MSI's 0.1% advantage in the benchmarks barely puts the X79A-GD45 8D over Intel's DX79TO for the top two positions.

Overclocking

The TPower X79's fixed and offset Vcore settings can be used simultaneously, boosting its theoretical maximum to 2.06 volts.

Though several of this round-up’s motherboards offer base clock increments of less than 1 MHz, the best overclockers weren’t among them. ASRock’s X79 Extreme4 and MSI’s X79A-GD45 both reach 4.70 GHz. Neither could push our processor to 4747 MHz (which would have been required for 47 x 101 MHz), and the lack of more granular base clock settings results in a tie.

ECS gets excluded from this overclocking comparison because its old firmware did not support our newer C2-stepping CPU. Comparing a C1 overclock to a C2 wouldn't yield a fair result, especially since our C1-stepping chip is afflicted with a bug that prevents proper operation above a certain multiplier (and only Asus has been able to work around it). The new CPU is also less capable of supporting a 1.66x chipset-to-CPU multiplier, which puts the boards employing it at a disadvantage.

Every Sandy Bridge-E processor has a frequency window that includes an underclocking floor. ASRock’s 152 MHz base clock, which is achieved using a 1.66x strap, requires the base frequency be underclocked to 91 MHz, which is roughly the lowest setting this CPU supports.

The 1.25x strap is far easier to use, and the top three boards to employ it all reach 134 MHz. This capability is particularly useful to owners of locked processors like the Core i7-3820, since they’re still be able to increase their CPU’s frequency up to 34% beyond its highest multiplier.

Asus does wonderful things with memory overclocking, which is why we often use its products as our reference platform in memory reviews. Intel follows closely behind, though it isn’t able to effectively use its DDR3-2133 setting when all eight slots are filled. Pushing past the possible DDR3-2400 setting would require far-more-aggressive voltage adjustments than we’re comfortable using.

Which X79-Based Motherboard Offers More Value?

Today’s motherboards range from Intel’s $210 DX79TO to a $260 price shared by Asus, MSI, and ECS. Gigabyte’s X79A-UD3 might have also fit within our qualifying range, had its board not already been used for our $260 to $320 roundup. If you want to know more about the Gigabyte platform, check out Seven $260-$320 X79 Express Motherboards, Reviewed.

Because all of today’s boards exhibit similar performance, a price-per-performance chart primarily reflects price.

For $5 more than Intel's baseline, ASRock’s X79 Extreme4 adds three-way SLI, front-panel USB 3.0, rear-panel digital ports, twice as many analog audio jacks, three more SATA ports, and eSATA. While all of those features are worth several times as much as the price difference, we're left wondering how much money ASRock saved by leaving out the second set of DIMM slots.

Biostar’s TPower X79 goes a step further by integrating four more USB 3.0 ports compared to the X79 Extreme4. It still has only a single set of memory slots, however, and we’re unhappy to find its front-panel USB 3.0 header located less than one space below its third graphics slot. While most builders at this budget level don’t plan to use three graphics cards, we like to leave that option available when the slots are there. Furthermore, the TPower X79 is an extraordinarily troublesome overclocker. All of these things diminish the value of its added peripheral connectivity, compelling us to choose between ASRock and Intel.

Asus’ P9X79 splits the difference between the two most value-oriented boards by offering four rear-panel USB 3.0 ports but—strangely—no front-panel USB 3.0 headers. The board also comes up two SATA ports shy of the X79 Extreme4, and its super-valuable USB BIOS Flashback feature is so infrequently needed that it alone can’t justify its significantly-higher $260 price.

Nobody needs a win right now more than MSI, and the firm does beat its competitors with impressive efficiency. Like the TPower X79, MSI’s X79A-GD45 also has the problem of an unfortunately-placed front-panel USB 3.0 header. Anyone who knows for certain that they’ll never place a conflicting card in the bottom PCI Express slot should consider its superb efficiency as a reason to buy, though we find it hard to justify a $45 mark-up (compared to the similarly-featured X79 Extreme4) for its second set of memory slots.

ECS’ X79R-AX Deluxe faces several challenges compared to the competition, beginning with the fact that it was the only board to require a firmware update simply to support our new CPU. Anyone who orders a new processor with this board could potentially suffer the same fate. This is one of those instances where a feature similar to Asus’ USB BIOS Flashback could truly be useful, but ECS doesn’t have it. That could explain why the X79R-AX Deluxe is out-of-stock everywhere we looked.

This leads us back to a race between ASRock’s X79 Extreme4’s added features and the extra memory slots provided by Intel’s DX79TO. We could try to use the DX79TO’s higher efficiency to offset its lack of front-panel USB 3.0 support, but we prefer to have both. We were also disappointed that Intel chose to use only 32 of the CPU's 40 PCIe lanes by leaving off the third graphics slot. Add superior CPU overclocking to its superior feature set, and ASRock’s X79 Extreme4 takes the value crown.

Following the initial disclosure of Ivy Bridge at IDF Fall 2011, Intel engineer Scott Siers announced that there will be four different Ivy Bridge die models. The dies will integrate two or four cores, two different DX11 graphics units, as well 2 to 8 MB L3 cache. Ivy Bridge will carry up to 1.4 billion transistors that span over an area of 160 mm², which is about 26 percent smaller than the comparable 216 mm² Sandy Bridge die with 1.16 billion transistors.

Ivy Bridge will also integrate DisplayPort support and 20 channels of PCIe 3. The memory controller now supports 1.35V DDR3L SODIMMs.

Digitimes reported last week that Ivy Bridge would not be shipping as planned, as the production of the processors appears to be delayed. According to the publication, a small volume of Ivy Bridge chips will be available in April with volume shipments not happening until Q3, which would mean - if Digitimes is right - that volume Ivy Bridge computers won't be available until early Q4. The culprit apparently is the economy and weak demand, which makes it difficult for PC makers to get rid of their Sandy Bridge inventory. However, Intel told VR-Zone that the report is only partially true and only dual-core Ivy Bridge models will be delayed.

The X-Lab Unlocked

Meet Pete. Pete Cibula, Jr. has a job that most geeks would kill for. His daytime existence is spent making sure that when you want fast service from your cloud service provider (whether that’s OnLive, Amazon, or Google), the speed is there. When you want to transfer half of a terabyte over your LAN, Pete’s work is part of why that process doesn’t take several days. Cars, tanks, slot machines, space shuttles, TVs, and plenty more all share the common thread of Ethernet. Pete’s job is to make sure that when the world needs to move beyond 1 Gb/s this year, the bandwidth will be there, the ports will work, the cabling won’t crush data streams, and the world’s communications will continue to advance.

Pete isn’t alone, obviously. Several manufacturers work in the networking silicon space. But Intel has a position of prominence and legacy that is unique. Most of us hardly give a second thought to Intel’s role in Ethernet development, but the efforts of Pete and his colleagues are literally helping to shape the networking experiences every one of us will enjoy and depend on in the near future.

The work they do happens in several rooms of the Jones Farm 3 building at one of Intel’s Hillsboro, OR campuses. As in our prior Western Digital venture, Tom's Hardware teamed up with pro photographer Gary Wilson to explore and reveal the little-known world of…Intel’s X-Lab.

The Need For Speed

The story behind the X-Lab isn’t just about the race to 10GBASE-T or making sure that when every server or workstation that needs more than one gigabit port, there’s a device able to reliably supply that bandwidth. The X-Lab exists to pave the way for the future of networking. In more than one way, networking scales alongside of CPUs and Moore’s Law. As systems are able to process more data, they need to exchange those greater data loads more quickly in order to maintain real-time functionality. Also consider the role of virtualization in networking. As companies condense five or ten servers into one physical machine, the networking load of those old servers gets jammed into a single box. What might have been  2-4 Gb/s of bandwidth spread across each of ten systems now has to flow in and out of only one, and this requires some fresh infrastructure changes.

Since 1978, Intel has led the industry through multiple speed transitions, from 10 megabit Ethernet through 100 megabit and gigabit Ethernet and now on to 10 gigabit Ethernet (10GbE). The IEEE may be responsible for the creation and supervision of the underlying specifications, but someone has to get their hands dirty and do the years of costly hardware development and validation. A huge chunk of that work gets done by Intel.

X-Lab Inside

Pete (right) and Joe Edwards (left) guided us into Jones Farm 3, past security, and through the  labyrinthine nest of laboratories and cubicle farms that is home to Intel’s LAN Access Division (LAD).

The LAD is a global operation, with silicon design centers in Israel and Austin, Texas; software development in Oregon and Poland; network interface card design and operations in Oregon; and several design centers in Asia and Europe. Knowing the magnitude of the work done here, and also knowing that the X-Lab is the epicenter of Intel’s 10GBASE-T networking platform test and validation efforts, I went in expecting something like a data center—some expansive room with raised flooring and rack after rack of test equipment. What I encountered was something closer to the equipment closet in the hall outside that data center.

The X-Lab, as Pete described it, is a “compact but efficient command-and-control center for twisted-pair Ethernet conformance testing.” Half of the chamber was dominated by chrome baker’s racks laden with what seemed an endless supply of Ethernet cabling. There were racks supporting test equipment and ever more bundles of cables. But I’d be stunned if the room was larger than 1000 square feet. Six of us standing amidst the benches and tools were rubbing elbows and finding it awkward to maneuver. Remodeling to expand the LAD’s test capabilities was going on in the adjacent room, so our conversations were constantly punctuated with the sporadic rhythm of hammering.

Yet, even in those cramped quarters, geek humor prevails. Pete has a Guy Fawkes mask perched atop one instrument, and a crown from some buffalo wing establishment adorns another shelf. Some years ago, one of Pete’s young children drew him a poster-sized landscape in crayon, scrawled unabashedly in green and yellow. The X-Lab crew still have it taped to the inside of the lab’s only door. They shrug with acceptance of their long hours and joke that it’s the most sunshine they usually see in a day.

Big Effort, Little Chip

If our tour was any indication, this is the current big Kahuna in the X-Lab: Intel’s 40 nm, 10GbE Twinville silicon. We photographed it here in both its LAN-on-motherboard (LOM) and PCI Express network interface (NIC) card (code-named Twinville) incarnations. As we proceed, you’ll begin to get a sense of the scale and resources Intel is devoting to making 10GbE happen.

“Ethernet is important for all servers, PCs and workstations, and it is the critical backbone of today’s datacenters,” said Pete. “That’s one of the reasons Intel reorganized in 2009 to combine Intel’s networking, server, and storage groups into a single data center group. 10GbE has already been deployed in many data centers.

Ethernet can be deployed over different types of physical interconnects. When a new, faster version of Ethernet is introduced, it starts out using expensive fiber optic connections. Eventually, as costs and power decline, it moves to the less expensive, copper interconnects the industry refers to as ‘BASE-T.’ These BASE-T connections use the familiar RJ-45 connector, which looks like a phone plug, only a little larger. This is the connection you see on nearly all servers, desktops, and laptops or in the Ethernet switches scattered throughout your favorite LAN party.”

To succeed in the market, 10GBASE-T needs low-cost cabling, run lengths of up to 100 meters, and backwards compatibility with gigabit Ethernet networks. The amount of silicon development, interface testing, power analysis, software validation, and everything else required to bring a new networking platform into production is almost overwhelming.

10GBASE-T...And Beyond

Twinville is Intel’s next-generation 10GBASE-T product, originally shown for the first time at the Intel Developer Forum last September. It will be the industry’s first single-chip, fully integrated, dual-port 10GBASE-T controller. Why does that matter? Because until now, 10GBASE-T designs have included at least two chips and a myriad of support components. This single-chip design will allow 10GBASE-T solutions to finally be small, cheap, and power-efficient enough to be integrated onto server motherboards. Expect this to happen in 2011.

Let’s take it a step further. Why do we specifically need 10GBASE-T? Because today, if you want to go faster than gigabit Ethernet, 10GBASE-T offers the easiest upgrade path. Other flavors of 10GbE and different network fabrics require major infrastructure changes (new switches, cabling, and so on). 10GBASE-T’s backward compatibility with existing GbE networks means you can install 10GbE adapters that will work with your existing equipment, and when you’re ready to upgrade switches and cabling, the move is easy and seamless. In fact, the transition is so smooth that the Dell’Oro Group predicts 10GBASE-T port shipments will grow from 5 million units in 2011 to more than 25 million by 2014.

By the time 10GbE becomes that prominent, the world will start to need terabit Ethernet switches, and we’ll be repeating the same situation we have today with another zero added to each number. Fortunately, the just-released 40GbE and 100GbE standards will help ease the transition.

NIC Evolution

“Twinville will be used on our fourth-generation 10GBASE-T adapter product,” said Pete. “Our first-generation card here at the top was a single-port, 10GBASE-T adapter. It used the 82598, had a third-party PHY, and it came in right at the top of the 25 W limit for PCI Express. The PHY itself burned approximately 14 W of that power budget and required an active cooling solution. That product was introduced in 2007. We then ‘upgraded’ that 82598 MAC by coupling it with a second-generation, 65 nm 10GBASE-T PHY. With that single-port solution, we were able to lose the active cooling. Losing active cooling is critical for LAN-on-motherboard solutions. You don’t want more fans…although blue LEDs might be nice! Anyway, we called this the WWF heatsink because it almost looks like the WWF logo. That design gave us enough surface area to dissipate the heat without a fan, and the total power was significantly lower—about 16 W.

Then, using a similar 65 nm device, we were able to use the same active [PHY] heatsink but go to a dual-port solution. Total power bumped up to approximately 20 W, which is still lower less power than the first-generation device, but with two ports. It uses the 82599 media access controller and has less board complexity. There are fewer power components, for example.

And now, moving to our 40 nm device, Twinville, you can see that the MAC is integrated with the PHY. Actually, it’s not just one physical interface—it’s two. Also, it supports three speeds—100 Mb, 1 G, and 10 G—whereas the others only supported two speeds. And it’s expected to consume about 10 W. The Twinville we have in testing now uses an active heatsink, but by production it’ll be passive, just like the second-gen card.”

We Like To Conform

Pete struck me as a humble guy and a true team player. In the several hours I spent with him, he only once mentioned a personal accomplishment: “My one claim to fame is probably that I’ve figured out how to automate a lot of this signal testing. If it’s not automated, it’s very time-consuming.”

One of the many signal tests done in the X-Lab involves physical layer transmitter conformance. This characterizes the transmitted signal quality, or, as Pete put it, “makes sure that the signals sent out on the wire are wiggling correctly.” Intel tests these parameters using a mix of equipment, including RF power meters, RF spectrum analyzers, RF vector network analyzers, and oscilloscopes. Quite often, the X-Lab team has had to develop its own custom software tools to assure accurate and repeatable test results.

With the lab’s automated test implementation, a basic 10 Mb/100 Mb/1 Gb test pass can be completed in about six hours. Only about 30 minutes of this requires hands-on operator interaction. Using the prior, manual test bench methods, the same test pass would take about two weeks for 10 Mb and 100 Mb alone. Similar time savings are realized in the lab’s 10GBASE-T testing.

Signal Analysis

Wrist-deep in Twin Pond with Twinville A1 silicon stepping, we moved through some signal integrity measurements. Specifically, we examined transmitter droop using an oscilloscope, which shows waveform decay over time. Next, we switched to the spectrum analyzer. At this point, Pete stopped in his tracks, realizing that he’d forgotten to lecture on the channel requirements for 10GBASE-T and how those played into the change in instrumentation.

“CAT6a has a nominal channel bandwidth of 500 MHz. Some manufacturers extend that to 650 MHz,” he explained. “So, it’s a very broad range. The signal encoding is so complex that it’s very difficult to make a measurement on a time domain signal. To address that issue, the IEEE specifies a series of test measurements that are defined more in the RF realm than in the time domain. Right now, we’re selecting one of the four twisted pairs, routing it into the spectrum analyzer, which is acquiring the signal for us, and then we do some offline analysis of the signal. In this case, the specific measurement is called spurious free dynamic range. We’re using the analyzer to acquire a two-tone test signal and we’re looking at differences between the signal’s minimum and maximum frequency peaks. We do that with multiple pairs of tones, and taken together, the measurements define the linearity of the system.” 

Hot And Cold

Networks have to function everywhere—from the frigid Subarctic to the enclosed swelter of an engine room. X-Lab techs call this Thermonics temperature forcing device the “Elephant Arm,” and it can drive component temperatures down under -55˚C and up to over 250˚C. The idea is to focus an air stream onto the case of a device.

Techs needed extreme cooling when working on first-generation devices owing to their 17 W to 25 W power envelopes. The amount of heat generated under load required significant air conditioning to get the chip down to 0ºC. Pete would have demonstrated, but the machine is loud enough to require ear protection, and we didn’t have any on hand.

CAT Fight

CAT5e cabling is very common. I have it running through my own walls for gigabit structured wiring. The cabling specs out at 125 MHz and simply uses four pairs of loosely twisted copper wiring. CAT6 steps up to 250 MHz, features tighter twisting, and adds in a dielectric conduit that separates the four twisted pairs and helps prevent energy from one pair bleeding into its neighbors.  Most 10GBASE-T specifically targets CAT6a cabling, which includes even more stringent control over pair twist, as well as manufacturer-specific design features for improving immunity to alien crosstalk noise.

Pete set 10GBASE-T cabling in historical perspective to illustrate some of the increasing problems networking engineers face.

“10BASE-T has been around for a long time, and it’s even been demonstrated to work over barbed wire. There’s that much signal-to-noise ratio margin, even on a very lossy channel like barbed wire. It’s 6 V peak to peak, best-case, and the pulses are very wide. It takes a lot of bad things happening in the channel for a receiver not to see it.

100BASE-TX requires you to do some funky things to the signal. Instead of the two-level signal in 10BASE-T, there’s a three-level signal. If you look at the signal energy, it’s a 2 V peak to peak system, so there’s less power, but all of this scrambling and pulse shaping gets it to work. In 1998, that all wasn’t very straightforward, but with modern signal processing, it’s pretty easy to do 100BASE-TX.

Now, for gigabit, you start to get into some magic. There is more noise power than signal power, meaning we have a negative signal-to-noise ratio. That means if there’s a lot of background noise—like in this room now—and if we get that hammering noise so loud that you can’t hear me, that’s like a gigabit Ethernet noise environment. In 2000, 2001, there were some signal processing techniques applied, some special encoding and decoding that, at a high level, means you’re taking a best guess. You know what you’re sending out, and the receiver knows that there are certain expected combinations that will be coming back. So the system takes a best guess, to put it crudely, at what that data is. Better than 1 in 10^-10 times, it makes the right guess.

But gigabit sucked up a lot of power and required, for the time, a lot of gates. At the gigabit inflection point, you started to have more gates than analog circuitry because we’re sending highly-encoded analog signals—those wiggly things with an amplitude and everything else. To encode and decode that properly requires a lot of logic gates. For 10GBASE-T, you just carry that concept to the next level. If gigabit is a whisper in a rock concert, 10GBASE-T would be like a whisper in a nuclear blast. It’s that much more noise power compared to the signal power. But with today’s digital signal processing techniques, you can make a signal have more apparent power. That’s one way to think of it. Again, the ratio of analog content to gates in 10GBASE-T is—wow. It’s very significant, with much, much more digital than analog content. This is good because it suddenly becomes very Moore’s Law-friendly, plus you get the advantage of power savings as you go to each new process node. In the lab here, we have 90, 60, and 40 nm technologies. The power savings associated with each generation has been key for our 10 Gb NIC products and the broader 10GBASE-T deployment in servers and switches.”

Miles Of Cabling

A lot of the X-Lab’s testing is done in different ways over various segments of cable. The lab contains cabling from seven manufacturers spanning 14 cable types. Pete said that he stopped doing the math at seven miles and two tons of cable, “just because…it’s a lot.”

Intel has to look at so many varieties because they’re constructed differently, and so have different characteristics. The group needs to ensure that its products will work over whatever is commonly installed. “We don’t have every manufacturer in the world in here,” said Pete, “but we asked the cable distributors who is the most popular and widely deployed, and that’s what we got.”

As you wonder at all of that cabling, think about your feet. In particular, think about your feet dragging across a carpet, then touching the light switch with your fingertip. Friction followed by static electricity buildup and discharge, right? Now imaging snaking those miles of cables through walls and crawlspaces, the friction causing electrostatic buildup on the cabling jackets. What do you suppose might happen when someone goes to plug that cable into a patch panel? Maybe nothing…or maybe not.

“I was a cable discharge skeptic until we actually pulled cables into the lab and I got zapped by a cable, “ says Pete.” Guys were working next door to us, pulling cable through conduits and onto racks and trays. So you pull the cable, you plug it into a port—well, the switch is going to provide a passive ground for that charge. We’re verifying the immunity of a networking device—a port—to that type of a discharge. It’s different than the typical ESD testing we do anywhere else in Intel or at any semiconductor company. Those involve looking at the movement of people, like across a carpet, or the movement of machines building up a charge as they operate. But this is a special type of ESD that really only appears in the networking world.”

The Lightning Bolt

The IEEE formed an ad-hoc committee in 2001 to investigate cabling ESD, but the post-dot-com recession doomed the group’s efforts. The X-Lab’s current testing methods were salvaged from that aborted effort. Pete and company built a special cable discharge event tester called Thor (because Zeus was already taken by another machine) to replicate the type of shock that might manifest in a bad cable shock.

Actually, while Thor can charge up to 5000 V, air discharge arcs over at about 3000 V within a patch panel, leaving a maximum practical charge of 2000 V. That’s still enough to damage a server, switch ports, or other equipment. Results observed in a Thor discharge will show whether tested adapters will survive those shocks. Before our visit, I’d joked with the Intel team that it was always fun to watch stuff blow up. Pete remarked that this might be our chance.

Thor Strikes!

We installed a pair of wires into a patch panel and used current and voltage probes to measure the discharge. We cycled through three tests—500 V, 1000 V, and 2000 V—using an oscilloscope to record the data and channel it into Pete’s laptop, which he had me operate. “Be careful,” he warned me on my way to the bench, not really able to hide his smile. “We blew up a connection doing this not long ago.”

I won’t mislead you. Nothing blew up. I got to see some heavy duty spikes in the scope readings, but our connections and equipment remained intact. On my way to sit at the bench, however, I did crack my head pretty good on Thor’s bottom edge. Electrically safe, yes, but let the unwary beware.

The Rack Of Doom

Where there’s transmitter conformance testing, expect similar treatment for receivers. For this pupose, the X-Lab has its so-called “Rack of Doom,” shown here. Tests using the equipment in this rack ensure that the receiver meets the receive error rate requirements defined in the IEEE Ethernet specification. For 10 Gb technologies (including 10GBASE-T), the IEEE defines a bit error ratio (BER) target of no more than one error in every 10¹² bits received. That’s one error in every trillion bits. And if that’s not impressive enough, a 10 Gb receiver must perform with this same precision even when bombarded with various types of noise or when plugged into an erroneously wired connection, which isn’t uncommon in places where techs custom-cut their own cabling.

The Rack of Doom contains multiple random noise sources, some of which get used in alien crosstalk noise testing (see below). This ensures that the PHY’s internal digital signal processing (DSP) noise-cancelling systems function properly in the presence of noise that they don’t understand and therefore can’t cancel. In 10GBASE-T communications, this usually manifests as signals bleeding into the wire from adjacent cables in a cable tray.

Enter Noise

Data communications have three necessary components: a transmitter, a receiver, and something in between, even if that something is air (as in Wi-Fi). The medium is called a channel, and channels are partly defined in terms of electrical characteristics, including signal attenuation along the length of the channel. The longer the cable, the more power declines along its length. This is one problem with which the X-Lab has to grapple.

Techs are also concerned about echo. Every interconnect point along the channel, such as a switch, creates something akin to a bump in the wire. At these points, part of the signal’s energy gets reflected back. In fluid dynamics, this would be like going from a big hose to a little hose and dealing with eddies and turbulence. The echoed signal returning to the transmitter distorts the main signal flow. Echo is one of several noise types that X-Lab staff must tackle. I was about to find that there were plenty more.

Six Around One

Pete noted, “Another thing that happens when you have multiple signals in a cable—and we have four signal pairs here—is crosstalk. This is when signals from one wire couple onto another wire. There are various forms of crosstalk. Within a cable, there are different levels of crosstalk, or XT, energy at the near end of the cable than the far end. We call that NEXT and FEXT. You can also get crosstalk from one cable to another—that’s the infamous alien crosstalk.

Alien crosstalk is noise introduced from an adjacent channel. One way we introduce that is to intentionally create a worst-case scenario. See the cables we’ve bundled and tightly wrapped around this rack? There’s a red cable in the bundle called the victim, and there are six aggressors, or disturbers, around it. We send traffic over the six pairs around the one. Colloquially, we call it a ‘six around one’ test. We try to couple as much alien energy into that cable in the center as possible and see how well we do. This isn’t designed in the IEEE standard, but it’s more of a real-world type test.”

Spendy Noise

“Let’s take that noise problem one step further,” said Pete, drawing me over to a nearby bench to examine a small PCB with a dissected Ethernet cable tied across its middle. “Say we run these alien crosstalk tests. We see errors. The next thing we want is to see what energy is getting coupled in there. So you get an oscilloscope or spectrum analyzer or some piece of equipment to measure the noise on that center channel while these other guys are sending stuff. You can mathematically model that noise and define it in terms of some type of more standard noise source.

In the standard, the IEEE has defined a white Gaussian noise—think of it as random noise—that is flat across the frequency range from 10 to 400 MHz with a specific noise power of 141.9 dBm/Hz.” I raised an eyebrow at the specificity, and he shrugged. “Hey, I live with this and don’t get out much. But anyway, the IEEE says you want to grab four independent noise sources, which we have here, and couple those into each pair of the wire. While you have that noise coupled in, see how well you do. It’s like somebody’s yelling at you and you’re still trying to communicate.

We’ve got, in this case, one fixture that we’ve designed ourselves using some very precision noise sources that are quite spendy. How much? Each one of these little things is $2000. This is $8000 of noise here. We tend to think of things in terms of equivalent values here. ‘This is like a used car. That thing over there is like a house.’ With truly random noise, you cannot mathematically cancel it. Instead, you just have to figure out ways to deal with it.”

Whither The 1.21 Jigowatts?

Those who have seen Back to the Future no doubt recall Doc and his revelation about the DeLorean’s “1.21 jigowatt” power requirements. Thankfully, we had neither plutonium nor lightning strikes present on our tour, but we did have a flux capacitor. That’s the cylindrical capacitor housed within the bisected metal block you see above. The data cable sets inside the center conductor, and it literally is in flux—hence the name. In electromagnetic field theory, there are lines of flux, so X-Lab engineers joke about it being the infamous “flux capacitor.” In reality, all capacitors are flux capacitors. Gotta love that geek humor. Anyway, the inserted cable has noise coupled onto all four of its wire pairs simultaneously.

Techs examine several signal types through the flux capacitor test to assess their effect on differential signaling, which assists with noise cancelation.

“Differential signaling is a very brute force method of noise cancellation,” explained Pete. “We split the signal, invert one half, run this split signal into a difference amplifier, and look at the resulting signal. If you have a signal that is common to both halves—meaning the same phase, polarity, and amplitude—coming into the receiver, the signals should cancel. Well, that’s great in math, but in the real world, one pair of wire will probably be a little longer or shorter than another, so you have an offset in time. They’ll probably be manufactured slightly differently, so the amplitude or power will be different. You subtract two things that are not identical but close. That’s why you get a little noise left over at the end.”

I asked if this was the same principle behind active noise cancelation in headphones.

“That’s a good way to think about it,” he said. “By the way, our alien crosstalk test uses a differential noise source. It comes in and stays there. For common-mode noise, we’re hoping that the system will work really well, the noise will balance, and be eliminated. But in the real world, it doesn’t always happen.”

(Note: For those who question the spelling of “jigowatts,” see http://www.imsdb.com/scripts/Back-to-the-Future.pdf at the famous shot 67.)

Cell Phones, Always Interrupting

Many parts of the IEEE standard are “normative,” meaning they must be followed and adhered to. But a good manufacturer doesn’t want to merely meet spec; it wants to exceed it and see how far performance will go until the system breaks.

“For first-gen adapters, it wasn’t that hard to make them fail,” said Pete. “People would walk in with old 900 MHz cell phones and you’d see things drop in a heartbeat. That’s why we added this part called, oddly enough, the cell phone filter. It’s an additional passive filter network inserted between the magnetics module and the silicon chip. It’s inserted into the circuit to filter out those frequencies. So, with the first generation, we just put a passive filter on it. And it helped a lot. Otherwise, we wouldn’t have been able to ship a product.

“But additionally, you can design into algorithms the firmware to detect interfering radio signals and compensate for them, because you have walkie-talkies, cordless phones, and everything else. So the second-generation adapters included both the discrete cell phone filter and those algorithms. We’re continuing to raise the bar, too. I mean, what if someone wants to bring a radar system into the lab? OK, maybe that’s an exaggeration, but that’s the thought process. What are the potential scenarios? There are additional mechanisms I can’t even describe, because they’re proprietary, that enhance the noise immunity. But in the test environment, we have to crank up the volume of our disturbers to see where there’s the break.”

Power and noise immunity are “the last frontiers” for 10GBASE-T, and Pete proudly noted that their study and application is literally grounded in rocket science. Many of the principals and CTOs in the PHY companies with which Intel works have backgrounds in satellite communications. That’s where much of the theory behind 10GBASE-T technology has its roots.

“Those algorithms are how we hear a whisper in a thermonuclear explosion,” Pete added. “And hey! We have alien crosstalk. That’s where the aliens meet our rocket science!”

From Aliens To Robots

As mentioned earlier, the bit error ratio target for 10GBASE-T is one in one trillion bits. Confirming that rate (or better) requires multiple kinds of tests, but the X-Lab’s standard test pass involves 500 million packets using 1518-byte frames transmitted for approximately 15 minutes.  One hundred of these test passes then comprise a single test trial, and there are many such trials during a product test cycle.

“The reason we do that—thousands of link attempts and hundreds of millions of packets—is that 10GBASE-T is a huge math problem that has to be solved in real-time,” said Pete. “You don’t always get the same answer, because it’s non-deterministic. Because of that non-determinism, you really want to do a lot of trials to make sure that the system is converging consistently over these different math problems. Note than the full test trial is executed on one cable type from one manufacturer on a two-meter length…then on a four-meter, on a seven, on a 20, on a 55, on a 90, and on a 100-meter channel. Finally, the entire set is repeated with different manufacturers. You can see that we easily end up transmitting trillions of packets during the qualification phase.”

To enable the kind of 24x7 testing necessary when pushing all of those packets during a product test cycle, the Intel team has developed several robotic solutions to automatically configure test channels. One of these is a third-generation robot called Patch that was two years in the making. If you’re expecting Johnny 5 or Asimo, you might be a bit disappointed, but Patch is still pretty freaking cool up close.

Patch is about six feet tall and three feet wide, comprised mainly of a frame, a LAN jack patch panel, and a couple of plugging/unplugging devices that slide along a shelf in front of that panel. Patch is tethered to a control PC and tasked with plugging and unplugging network cables from its patch panel…over…and over.

Patch And Ports

Randy Kumenius, the engineer who designed much of Patch and often operates it, described the robot as “a two-axis robotic solution based on a four-axis National Instruments motion card. Both axes are closed-loop servo motors. We drive an analog voltage out to the motor, which sends back a four-bit encoder. We get 4096 counts per revolution because of the four-line encoder. So we can get very precise positioning. And it’s absolute positioning information, so once this one’s home and the next one’s home, we zero that and all of the ports are X amount of counts from zero.”

Patch can handle two cables of any length simultaneously, helping to double its potential work output and increase efficiency. Testers can analyze PHY communications in any length from one to 100 meters in one-meter increments. Patch has proven so useful and reliable as part of Intel’s Oregon site’s that its offspring will be deployed to multiple platform test locations in 2011.

Power Delivery Testing

Moving on. One can’t help but admire the array of different homegrown systems scattered across Intel’s labs. Chassis side panels seem scorned, and not much attention gets paid to PC cable management since there are simply too many mods and a semi-constant need to change equipment. It’s all function over form.

Take the above setup as an example. It’s a collection of gear designed to test the power delivery network of Intel networking silicon. Engineers look to characterize the chip’s power consumption, power quality, and efficiency. With the inevitable rise of 10GbE products, power consumption and efficient power delivery will only become increasingly important.

“10GBASE-T in particular has provided a new challenge in that, with these early silicon generations at least, we no longer have the power margins of 1GbE,” noted Pete. Some may recall that early 1 Gb Ethernet chips were quite power-hungry, as well.  “Our power delivery testing ensures that our 10GbE products adhere to PCIe specification limits as well as meeting customer expectations.”

Hot Technology

For systems running at high temperatures, Intel needed to create an isolated environment in which engineers could drive the temperature of the card independently of the system. The other PC components are fairly irrelevant. They’re all off-the-shelf and qualified elsewhere. In this case, Intel just wanted to focus on the NIC.

Engineers created a homebrew enclosure in which compressed air gets piped in and exhausted out the back. The very first design was made from cardboard, but then progressed to Plexiglas. Today, the design is widely used across Intel’s labs.

The in-progress test shown here has this NIC baking at 57.8˚C during operation.

Helping Out Buyers

I’m learning that, in the world of R&D and engineering, there’s a fine line dividing the clean room from the garage. This little space is just big enough for two rack rows, but it’s where Intel does the bulk of its validation and troubleshooting work for OEMs—original equipment manufacturers that purchase Intel’s networking silicon for use in their own systems. You can see holes in the ceiling where prior infrastructure once ran. The cramped room is relatively loud from the thrumming of system fans. In this photo, you can see the end cap, directly facing the room’s one entrance, where staff display their assortment of radio, gaming, and other bumper stickers. A certificate from the University of New Hampshire’s  InterOperability Laboratory, something of a Mecca in the Ethernet validation world, dominates the eclectic collection.

Patching And Problems

Once silicon and NIC reference designs are finalized, Intel will send the products to OEMs, who in turn create their systems and return early samples back to Intel for analysis. Intel then compares those OEM products against proven reference board results. If the OEM data looks worse, an Intel technical marketing engineer will go over the OEM’s schematics and work to help the manufacturers improve their quality. Sometimes, engineers are even called on to do failure analysis in the field.

“I’m doing a laptop project now where there may be a cold solder joint problem,” said Intel’s Matt Wonson. “Like you desolder and resolder the network silicon, and all the problems go away.”

Matt works in new products testing, customer designs, and troubleshooting. The testing and troubleshooting work done here looks somewhat familiar to that done in the X-Lab. For example, the station shown above is for doing bit error rate testing. Techs can use relays to patch cables for lengths up to 160 meters. “As the test runs,” said Matt, “it finishes one cable length, then uses the relays to create a sort of virtual cable.”

Part Of The Bargain

“This station over here is doing some transmit testing,” continued Matt. “It’s a little more complex, with more stuff on it. We’ll have the oscilloscope hooked up to that, measuring. We’ve got different cable models and stuff on the board, so as each test runs, the board switches to the appropriate load for that test. The way that the racks work is we’ve got a controller PC with a GPIB controller card in it that talks to all of the different instruments. So when it’s getting a run—say you have to have a certain capture length—the controller PC tells the scope to capture this long of a record, maybe along with some screen grabs. It’s got pass/fail criteria, what’s marginal, what was actually measured. Right now, I’ve got these Powerville 4-port, GigE, no bridge chip network cards that do Energy Efficient Ethernet. That’s one of the big things in power right now.”

Intel doesn’t charge for this support. Anybody who buys Intel silicon can get these test results for their product. The University of New Hampshire has its Interoperability Test program, and it’s the most renowned lab in this space, but, according to my tour hosts, the consortium fees can add up fast for a low-margin manufacturer.

“So we’ll provide all that support plus review your design and suggest what components to use on magnetics or resistors or whatever,” said Matt. “We may not always be the lowest-cost solution, but we bring a lot to the table.”

Thinking Inside The Box

On the odds-and-ends shelf near the X-Lab’s door, there’s a small, clear box containing a palmful of trinkets. It reminds me of a similar jar of junk I have in my garage—some random memorabilia from my childhood. The box belongs to Pete, and I wondered if its contents might somehow illuminate his personality and perhaps his work at Intel.

“This is my way of remembering the moments,” said Pete, gazing into the box. “In my prior career, I was actually a failure analysis engineer. My job was to find problems with chips. Someone found this little bugaboo thing, a little digital IC thing with a capacitor. It looks kind of bug-ish. Over the years, I’ve thrown in different chips that have caused me great pain and excitement. And it was a place I could throw my chunk of trinitite. It changes over the years, but it’s been pretty stable because Intel is pretty good!”

Trinitite is fused glass from the first atomic bomb test in 1945. I have to assume that, for Pete, the shard represents a landmark shift in technology and its impact on the world.

“The most outstanding of these chips is probably this little motion control chip from my prior job,” he added. “We were starting to do some analysis on it, and the top literally fell right off in the middle of testing. That’s not something you want to have happen to your silicon.”

Our Token Ring Of Techs

What’s a good tour without a group shot? Pete, surrounded here by several of his teammates, holds some CAT6a cable and prepares to plug back into his day job, having finished shepherding us around his labs. Wrapping up, it’s clear to me how much we take for granted. We plug in our network connections, they work, we’re happy, end of story. Only it’s not. There’s a mind-boggling amount of work, ingenuity, and expense (the X-Lab room alone houses over $5 million in test gear) that goes into making your network connection possible and reliable.

Pete and his team are a vital part of making your data connection a reality. And when you get faster feeds tomorrow, when 10GbE finally lands on your desktop in the next few years, and when you can transfer terabytes in the time it now takes you to send gigabytes, maybe give a silent nod to the X-Lab crew and remember all the years of work that made your connection possible.

While many consumers have used 802.11n wireless routers for around two years now, it was assumed that the 802.11n standard was finalized before now. However, that's not the case as verified by Bob Heile, the chairman of the IEEE 802.15 working group on Personal Area Networks. In a recent email, Heile confirmed that the IEEE 802.11n draft standard was finally submitted to the Standards Review Committee (RevCom).

"On other fronts, 802.11 was granted unconditional approval to forward 11n to RevCom," Heile wrote. "After a bit of a rocky period on getting acceptable coexistence language included in the draft, I was pleased to support this approval. Congratulations to Bruce for his patience and perseverance in getting this done. This was an extremely complex project."

According to PC Magazine, the road to this point has been a long one for the 802.11n standard, its evolution dating back to 2004. An early draft version was approved in January 2006 (1.0) that would eventually kick-start the first wave of routers implementing the draft-n standard. But when the draft 802.11n standard failed to pass in May 2006, the Wi-Fi Alliance eventually agreed to certify the draft-n--or rather pre-N, based on Draft 2.0--products in June 2007.

However, based on Heile's recent email, the 802.11n struggle is expected to come to a full close by September 11 instead of the predicted January 2010 publication date.

If chip fabrication continues on a shrinking trend with Intel expected to release details surround its 32nm chip fabrication technology at the IEEE International Electron Devices Meeting (IEDM) in San Francisco on December 15.

Representatives from Intel will present a paper to discuss how they produced a 291Mb SRAM memory array to test the process – a process which uses high-k and metal gate technologies. Using a 0.171-micron^2 cell size, the device has close to 2 billion transistors and array density of 4.2-Mbit^2 – it operated at 3.8GHz at 1.1 Volts.

Static RAM cells are what chip makers always build first when testing a new fabrication process – and according to Intel, the new process appears to be good. Intel has also claimed that its first 32nm processors, codenamed ‘Westmere’ are expected to debut in late 2009 – essentially be a die-shrink version of their new Nehalem architecture, due out late this quarter.

Intel is also not the only chip maker detailing its progress, however. AMD is claiming the smallest functional SRAM cell ever made at 22nm, high-k and metal gate with a density of 0.10micron^2.

From the sounds of things, we can expect to start seeing 32nm chips from Intel around the third or fourth quarter of 2009 and 22nm chips from AMD. Generally the smaller they get, the lower the voltage – thus lowering the heat. This are going to be pretty cool next year, no pun intended.

When is a standard not really standard? When it’s the proposed IEEE P1901 standard for powerline networking, which the IEEE Standards Association’s IEEE P1901 Working Group will once again take under consideration when it meets in San Francisco November 4 through 6.

Unlike the IEEE 802.11n Draft 2.0 standard for wireless networking, the proposed IEEE P1901 standard for powerline networking does not guarantee that products certified as compliant with the IEEE P1901 standard will work together.

When two of the several camps competing to define the 802.11n wireless networking standard merged their proposals, the resulting draft standard guaranteed that all 802.11n products would interoperate. So when two of the three camps (the HomePlug Powerline Alliance and Panasonic) competing to define the IEEE P1901 standard for powerline networking merged their proposals in October 2007, many outside the IEEE assumed interoperability was an essential element. It’s not.

As currently proposed, the standard defines one unified MAC layer, but two different physical layers (one PHY layer based on fast Fourier transform orthogonal frequency-division multiplexing, as advocated by the HomePlug Powerline Alliance, and another PHY layer based on Wavelet OFDM, as proposed by Panasonic). The proposed standard allows manufacturers to build devices that utilize either or both PHY layers; but if they choose one over the other, that device will function only with other IEEE P1901 devices designed to use the same PHY layer.

To be fair, Webster’s defines “standard” merely as “something considered by an authority or by general consent as a basis of comparison; an approved model.” According to that definition, there’s nothing wrong with the proposal that the working group has failed to ratify at its three previous meetings. But it would also be fair to say that consumers equate “standard” with “interoperable,” assuming that as long as they buy a product that adheres to a given standard, it’s going to work with any other product that makes the same claim.

Rob Ranck, president of the HomePlug Powerline Alliance, acknowledges that consumers will need guidance beyond the presence of the phrase “IEEE P1901” on a powerline-networking product’s box. “P1901 devices will need to include some certification labeling with the PHY type,” said Ranck, “to ensure that consumers can tell the difference in the case of devices that don’t support both PHY types.”

But Chano Gomez, vice-president of technology and strategic partnerships at DS2 (a company advocating yet a third technology for a powerline-networking standard, which the IEEE P1901 Working Group has repeatedly voted down) argues that the proposed IEEE P1901 isn’t really a standard at all. In fact, he maintains that both the PHY and MAC layers are incompatible. “The current proposal under vote includes two incompatible specifications for PHY and MAC,” said Gomez, “which in my opinion is the reason why many companies oppose it. This is a truly technical reason to vote against the proposal.”

Gomez went on to say that “standards bring two benefits for the industry: they ensure that all products work together, which is good for consumers, and they provide certainty to new silicon vendors that want to enter the market. With more vendors, you get more competition and eventually more innovation and lower prices—which is also good for consumers. This two PHY, two MAC proposal achieves none of that.”

Ranck argues that the perfect is the enemy of the good. “Although some have argued against this approach in favor of having complete interoperability,” said Ranck “the reality is that the powerline industry has been fragmented for many years and there is no guarantee that a consensus on interoperability will be reached in the near future. HomePlug’s objective is to support all efforts toward unifying the industry, and we think that taking a step forward in this fashion with IEEE P1901 is a good thing.”

Powerline networking is a good idea—when it works. While few would argue that hardwired Ethernet using Cat5e is the best solution, it’s not always possible to retrofit a home to use it. Wireless Ethernet is another good solution, but radio signals have difficulty penetrating some building materials, and they have limited range. Powerline networking has a place in the market, but the IEEE P1901 Working Group could learn a few things about gaining consumers’ confidence from their colleagues on the IEEE 802.11n Working Group.

The IEEE P1901 Working Group once again failed to reach a consensus on a draft standard for broadband-over-power-line networking at its most recent meeting in Madrid, Spain. The joint proposal put forward by Panasonic and the HomePlug Powerline Alliance, however, remains the odds-on favorite to win the next round of voting, scheduled for November 4 through 6 in San Francisco.

In an interview after the vote, we asked HomePlug Powerline Association (HPA) president Rob Ranck why he thought the proposal failed to pass muster with the working group.

“Well, some people voted against the proposal for technical reasons,” Ranck said, “and others voted against it for non-technical reasons. If a proposal doesn’t pass after two votes, the group returns to the down-selection process. At that time, new proposals can be put forward and proposals that were eliminated earlier can be brought back.” When we asked Ranck if he thought some of the working-group members’ “non-technical” reasons for voting against the Panasonic-HPA proposal were political in nature, he replied “Well, you used that word, I didn’t.”

If those dissenting members were hoping that the previously rejected proposal put forth by the Universal Powerline Association (UPA) would win, they were likely disappointed. The UPA’s competing proposal is based largely on power-line networking technology developed by the Spanish company DS2; and the P1901 working group did re-examine it during the down-selection process, but ultimately voted once again to eliminate it from consideration.

Asked if he was confident that the Panasonic-HomePlug proposal would be ratified during the next vote, Ranck replied “We’ve been close twice now, but confident is one of those words I’m hesitant to use. I am optimistic, I will say that. If we can get the people who are voting against it for technical reasons to support it, we’ll be okay. I’m not sure we’ll ever convince those members who oppose it for non-technical reasons.”

There are three elements to the power-line networking standard, which the IEEE P1901 working group refers to as “clusters.” The Access cluster refers to power-line networking outside the home, where electrical utilities use the power lines to for meter reading and to monitor the health of the electrical grid. Internet service providers can also use the electrical grid to deliver broadband Internet access (typically in rural areas beyond the reach of more conventional services, such as cable TV service and DSL). The In-Home cluster refers to power-line networks inside the home. And the Coexistence cluster governs how similar technologies can operate within the same environment without interfering with each other.

IEEE rules require that any proposed standard win approval from 75 percent of the working group, which is a very high hurdle (one that explains why the IEEE 802.11n wireless networking standard remains stuck in Draft 2.0 mode). In the Madrid vote, 69 percent of the IEEE P1901 working group voted in favor of the Panasonic/HomePlug proposal for the In-home cluster and 48 percent voted in favor of that group’s Access cluster proposal.

After those two proposals failed to win 75 percent majorities, however, 80 percent of voters abstained when it came time to ratify the Co-existence proposal put forth by the HomePlug Powerline Alliance and CEPCA (CEPCA is another consortium whose members include nearly all the Japanese consumer-electronics manufacturers, including Panasonic, Sony, and Hitachi).The abstentions were something of a forgone conclusion since there was no standard to coexist with.

Semiconductor manufacturer DS2 has announced new features for its Aitana 200 Mbit/sec. powerline networking chipset that enable the component to deliver transmission speeds four times faster than standard wireless networks.

Powerline networking devices enable a home’s existing electrical wires to carry data—including audio and video—as well as electrical power. This enables homeowners to deploy hard-wired Ethernet networks without the need to string new wires. Unlike wireless networks, however, there is currently no IEEE standard for powerline networks, which means there’s no guarantee that one powerline product will work with another.

Three consortiums are competing to have proposals adopted as the IEEE P1901 standard for broadband-over-powerline networking: The Universal Powerline Assocation (of which DS2 is the principal member), the HomePlug Powerline Alliance (Intel is a major player), and the Consumer Electronics Powerline Communication Alliance (whose membership includes Panasonic). HPA and CEPCA merged their proposals last year and won an early round of IEEE working group voting, dealing a significant blow to DS2 and the UPA. But the game isn’t over yet.

The IEEE working group is scheduled to meet again on September 29. If the HPA/CEPCA proposal fails to win 75 percent approval in this round of voting, DS2’s proposal could be brought back for further consideration.

On Sep. 20, 2008, the Institute of Electrical and Electronics Engineers (IEEE) will stage its prestigious annual Honors Ceremony. Most notable among the award recipients is Intel co-founder Gordon Moore.

According to the IEEE press release, Moore is receiving the IEEE Medal of Honor "for pioneering technical roles in integrated-circuit processing, and leadership in the development of MOS memory, the microprocessor computer and the semiconductor industry." The IEEE Medal of Honor is the highest recognition that an individual can receive from the institute, which is given for an outstanding contribution to, or career in, the IEEE fields of interest.

Gordon Moore is probably best known for his 1965 Electronics Magazine publication "Cramming more components onto integrated circuits," where he detailed his original observation that the number of components that can be fit on an integrated circuit at minimum cost per component doubled every year. Dubbed "Moore’s Law" by Dr. Carver Mead at the California Institute of Technology (Caltech), this observation was refined in 1975 so that the law stated that the number of components doubled every two years.

Misconception ensued when a former workmate stated that computer performance would double every 18 months, because the transistors were becoming faster. This is sometimes confused with Moore’s law, however, as Moore explained in an interview transcript (PDF warning) published on the Intel website in 2005, he never mentioned an 18 month time period in his observations.

Other notable events with which Moore was involved include co-founding the largest semiconductor company in the world, Intel, of which he is now Chairman Emeritus. Moore and his wife also donated $600 million to Caltech in 2001 for research funding and a further $200 million in 2007 to Caltech and the University of California to build the world’s largest optical telescope.

Piscataway (NJ) - IEEE 1394, better known under the brand names of Firewire and i.Link, will get a speed bump before the end of the year: The IEEE has approved the new IEEE 1394-2008 specification that provides support for a bandwidth of up to 3.2 Gb/s.

Firewire has come a long way. From the initial development by Apple in the late 1980s, to the technology’s completion in 1995 and surge in popularity in the early 2000s, the technology has become a serial bus interface common in Sony and Apple computers as well as a range of consumer electronics devices such as video cameras.

Most IEEE 1394 devices are still running on the S400 (400 Mb/s) specification despite the fact that S800 (800 Mb/s) was introduced with the IEEE 1394b spec in 2003. The problem with this spec was a different connector than the design that was used for S400. However, bilingual cables that are compatible with S400 and S800 ports are available.

The IEEE today announced that it formally approved the IEEE 1394-2008 spec, which will introduce support for S1600 (1.6 Gb/s) and S3200 (3.2 Gb/s) while offering full backwards compatibility with S400 and S800 ports. Down the road, it is expected that IEEE 1394 will scale up to 6.4 Gb/s.

Firewire and i.Link desperately need the upgrade in order to remain competitive with USB, which will receive an upgrade to 4.8 Gb/s in version 3.0.

The IEEE 1394-2008 spec will become available in October, according to the IEEE. The USB 3.0 spec is expected to be published by the end of this year.

Still on top, but for how long?

With all the commotion concerning Intel’s upcoming “mainstream” chipset, it’s easy to forget that the X48 Express continues to be the company’s top pick. With full support for PCI-Express 2.0 graphics, the X48 already has the most significant “new technology” trumpeted for its “next release.” But the future look even brighter for today’s high-end buyers, as the X48 offers twice the number of PCI-Express 2.0 lanes as the upcoming mainstream part. Furthermore, product maturity assures that builders can expect fewer bugs with the current range of X48 boards than they would from the initial release of next-generation parts.

The X48 Express has been around for a few months and even longer if we include the previous X38 incarnation, but that isn’t keeping manufacturers from updating their product lines. Several companies are now producing updated revisions of previously released models, while a few took extra time to “perfect” their product. Today’s new models come from both expected and unexpected high-end motherboard manufacturers, including ASRock, DFI, ECS and Intel itself.

We’ve already covered the basic features of the X48 Express in our first X48 Motherboard Comparison, so let’s see what makes the new models different.

ASRock X48TurboTwins-WiFi

With so much of ASRock’s focus being on high-value performance, few of us expected the company to use Intel’s latest high-end chipset which, after all, is more expensive but electronically identical to the earlier X38 Express. Yet, the company leapt past the earlier enthusiast part, its X48TurboTwins-WiFi, with the likely intention to entice buyers across its entire product line by producing a flagship model. Will its highest-end product to date impress us enough to have the desired brand-boosting effect?

Layout and Features

Two things that stand out about the X48TurboTwins-WiFi more than its odd name are the dual PCI-Express x16 slots and simplified chipset sink arrangement, both of which scream “value” compared to competing X48 models. We say “value” rather than “low cost” because it’s well known that the X48 doesn’t run hot, and that the third x16 slot found on higher-cost boards is nothing more than a version 1.0 slot with only four PCI-Express pathways.

Thus, ASRock cuts costs in places where higher-priced parts are mostly for show. We seriously doubt anyone would want to put a high-performance graphics card in a third slot that has only an eighth of the bandwidth of the other two slots, so eliminating it was an easy choice for ASRock. Whether or not the simplified chipset cooling is a similarly “zero-negative-effect” solution is answered in this article’s overclocking comparison.

The X48TurboTwins-WiFi is still packed with features, and one that stands out at the high end is its use of both DDR2 and DDR3 memory slots. A feature normally reserved for lower-cost products, enthusiasts on a middle budget can build their system today while carrying over the DDR2-800 memory from their old system, and upgrade to DDR3 whenever prices become more tolerable.

Two of the X48TurboTwins-WiFi’s more questionable design elements are the placement of its 24-pin ATX/EPS power connector and WiFi USB riser card header. By putting the main power lead between the rear panel ports and the Northbridge, ASRock forces builders to pull the cable over or around the CPU cooler. The WiFi card blocks one of only two PCI slots that would remain useable when two double-thick graphics cards are installed. Addressing either concern would have required ASRock to sacrifice other features, however, such as back-panel ports and DIMM slots.

ASRock also places the floppy cable header very inconveniently below the lowest PCI slot, which could result in a cabling nightmare for Windows XP users who want to use the chipset’s RAID or AHCI modes for SATA. Perhaps they believe users will forgo these advanced modes or use Windows Vista.

Another look at the X48TurboTwins-WiFi’s top corner reveals two Serial ATA connectors, but these are not ports. Rather, these are pass-through connections for rear-panel eSATA, and rely on internal ports located at the opposite corner of the board. We’re fairly certain that most users will not want to string cables corner-to-corner across the motherboard, so perhaps ASRock could have gotten back some of the previously-mentioned missing space by simply eliminating the fake eSATA ports.

Flipping the motherboard over reveals some ICs close to the CPU socket, which could interfere with the installation of some CPU cooler support plates.

Onboard Devices

ASRock endows the X48TurboTwins-WiFi with a complete controller set typically expected of high-end motherboards, but differs from the norm by using a non-SATA add-in disk controller. Intel chipsets haven’t supported Ultra ATA devices since the P965, and ASRock would have likely left the board devoid of any add-in controller had this not been the case.

The X48TurboTwins-WiFi port panel provides PS/2 keyboard and mouse ports, coaxial and optical digital audio outputs, IEEE-1394 FireWire, six USB 2.0 ports, a single gigabit network port and six analog audio jacks. Two internal SATA to eSATA pass-through connections are also found here, but enabling these requires some very messy internal cabling.

ASRock chose JMicron’s JMB368 Ultra-ATA controller, rather than one that supports both Ultra-ATA and SATA. It uses one of the chipset’s six unallocated PCI-Express pathways to provide optimal performance…typically to slow devices such as a DVD burner.

Realtek’s RTL8111C uses PCI-Express to provide full bi-directional bandwidth to a single Gigabit Ethernet port. The motherboard’s single PCI-Express x1 slot and two third-party controllers use up only three of the chipset’s six “spare” lanes.

ASRock uses VIA’s ancient 6308 series FireWire controller to provide two IEEE-1394 ports. At 133MB/s, its PCI interface is more than fast enough to provide full performance for two 50MB/s (400 megabit) connections.

ASRock is the only company we know of to use Realtek’s ALC890B audio codec, which the company rates at a respectable 110db signal-to-noise ratio.

ASRock continues to use the AW-GA800BT mini-USB card previously found in Asus WiFi products to provide 802.11g/b wireless networking. Realtek’s RTL8187L controller does the job nicely.

BIOS and Overclocking

The X48TurboTwins-WiFi’s frequency range looks a little ambitious, with a top setting of FSB-3200 (800 MHz clock) exceeding chipset capabilities by a relatively large margin. On the other hand, the company only offers DDR2-667 and DDR2-800 compatibility with automatic ratio adjustment, rather than with manual ratio selection.

At least all the DDR3 speeds were available, but once again choosing a ratio is done by picking the speed, based on stock FSB. We’re sure newbie CPU overclockers will choose the wrong setting, overclock the memory too far and be confused as to why their system doesn’t boot.

Most disconcerting of available BIOS settings are CPU, MCH and ICH voltage settings of “Low, Middle, High and Highest.” Not knowing what voltage we’re applying to an overclock instills hesitation and fear.

The CPU configuration menu appears rather basic, but the most important controls are present. Voltage levels are found in the “chipset” menu.

ASRock includes only the most familiar memory timing adjustments. Its DRAM frequency settings appear adequate when DDR3 is installed, but are far too restrictive when DDR2 is used.

Scrolling down reveals the voltage settings, with a global “GTLREF” adjustment rather than per-core settings

Software and Accessories

ASRock has a slightly different version of its OC Tuner application for each motherboard. Features for the X48TurboTwins-WiFi include system health and hardware monitors plus overclocking and voltage controls.

The System Health monitor also includes an adjustment for automatic CPU fan speed control.

CPU speed detection shows power savings settings in action, when enabled.

ASRock’s overclocking utility provides FSB, PCIe and CPU multiplier adjustments on one menu and voltage control on another.

In addition to its own overclocking utility, ASRock includes McAfee Virus Scan on its installation CD.

The X48TurboTwins-WiFi includes only four SATA data cables, which will get used up quickly by anyone looking to enable the rear panel “eSATA” connectors via internal bridging. On the other hand, ASRock is the only brand we know of to include a cable for connecting the motherboard’s internal S/P-DIF output to a graphics card, which adds HDMI audio capability to certain GeForce models.

Asus P5E3 Premium WiFi-AP @n (Update)

We already reviewed the P5E3 Premium WiFi-AP @n edition motherboard in our original X48 Motherboard Comparison, or at least we thought we had. Asus later revealed a completely new circuit board, trickily given the same “revision number.” Because the review samples were not the same as the retail units, we secured the later version for a retest.

What is most noticeably different about the P5E3 Premium WiFi-AP @n Edition retail samples is a missing PCI-Express x1 slot in the top position. Some of the capacitors have also been moved around, and we have no idea what changes might be beneath the surface.

We’re only focusing the P5E3 Premium WiFi-AP coverage on changes, but we did find one in BIOS: Asus has added separate GTL Reference voltage adjustments for each CPU core.

Readers not yet familiar with the P5E3 Premium WiFi-AP can look back to our previous review for details on the things that haven’t changed in the retail circuit board revision.

DFI Lanparty LT X48-T2R

DFI has earned itself an excellent reputation among “elite overclockers” by adding a broad range of BIOS settings to its Lanparty series motherboards. The company is never in a hurry to release new products, however, as it prefers to take its time tuning them to face off with the best its competitors can offer.

Layout and Features

The most striking visual characteristic of DFI’s Lanparty LT X48-T2R might be its green theme, but our eyes hop immediately to the eight-phase digital voltage regulator. We’ve had slightly better overclocking stability in the past with boards that used a similar design.

Also noteworthy is that DFI uses all seven slot positions to fill its X48-T2R with expansion options. But while the board does feature three PCI-Express x16 slots, only the top two have 16 pathways and PCI-Express 2.0 transfer modes. The third x16 slot has only a quarter of its pathways (four lanes), and its bandwidth is further halved by the use of PCI-Express 1.0 mode.

The top two PCI-Express slots are spaced adequately to support huge triple-thick graphics coolers, and typical CrossFire configurations of double-thick cards will still leave two PCI and one PCI-Express slot open for other cards.

Most cable connections are ideally located for use in traditional tower case designs, with the exception of the floppy cable header on the motherboard’s bottom edge. Though required for loading RAID or AHCI drivers during XP setup, DFI likely believes that anyone who uses that installation method will remove the drive when it’s no longer needed.

Of slightly less concern are the eight forward-facing Serial ATA connectors, which could be blocked by the hard drive cages of some cases. The X48-T2R is an enthusiast motherboard, and DFI likely believes that enthusiasts will choose their cases appropriately.

The Lanparty LT X48-T2R uses separate sinks for the Northbridge, Southbridge and voltage regulator, while the Northbridge sink ships uninstalled. The sink supports a 60mm x10mm cooling fan using wire clip retainers, but is large enough to operate without its own fan in systems that have other fans nearby. DFI includes sink paste and a spreader to ease installation, but users requiring additional cooling must purchase their fans separately.

The Lanparty LT X48-T2R’s lower front corner features a post code display, power and reset buttons that are perfect for bench top testing and diagnostics, in addition to the poorly placed floppy header, several fan connectors and three two-port USB cable headers.

Onboard Devices

Logic would dictate that DFI’s use of three PCI-Express devices would leave only a total of three of the X48 chipset’s six available PCI-Express lanes available to the X48-T2R’s x1 and third x16 PCI-Express slots. Thus, if all onboard devices and slots are used, the third x16 slot must drop to x1 or x2 transfer mode, which is hardly worthy of any graphics card.

DFI’s rear-panel port section features a huge hole for VRM cooling, in keeping with the Lanparty brand’s overclocking theme. Only the basic connectors are featured here, and the boards added internal SATA ports seem geared for cases that have front-panel eSATA connections.

The Lanparty LT X48-T2R gets its Ultra ATA and two of its SATA ports from JMicron’s popular JMB363 controller. This device uses a PCI-Express x1 connection for data transfers up to 250MB/s in each direction.

A Marvell 88E8053 controller serves one Gigabit Ethernet connection over its PCI-Express interface.

Meanwhile, an 88E8052 provides the second Gigabit Ethernet connection, again through PCI-Express. Notice that DFI also offers jumpers for hardware overclocking, which is an option we hadn’t seen in a while.

The VT6307 FireWire controller adds two IEEE-1394 ports via legacy PCI connection.

The Lanparty LT X48-T2R’s “onboard” audio comes by way of a riser card, which is supposed to reduce noise compared to typical onboard codec locations.

The Bernstein audio module uses Realtek’s ALC885 codec to modulate HD signals from the ICH9R Southbridge. It features separate 7.1-channel rear and two-channel front outputs for audio multitasking, such as running a VoIP application over a headset while playing a movie soundtrack through speakers.

BIOS and Overclocking

DFI provides an excellent range of voltage and timing adjustments in small increments, allowing each overclocker to shoot for the best possible performance his or her components can achieve. The CPU core voltage range does look a little conservative compared to other settings, but DFI also provides a “CPU VID Add” setting to increase the core up to 30% beyond the set voltage, in 0.23% increments.

DFI calls its overclocking menus “Genie BIOS.” The main menu includes basic clock and ratio settings for the CPU and RAM.

The voltage submenu includes VID control, the previously mentioned “VID Add” percentage; DRAM, chipset and per-die GTL reference voltage settings.

The DRAM timings menu has every setting that even the most avid overclocker could want, and those less interested in the more advanced settings will find individual “automatic” options.

Software and Accessories

Hardcore overclockers will always prefer BIOS settings or even jumpers, rather than use overclocking software. With limited enthusiasm for such programs in the X48-T2R’s target market, DFI simply included the third-party freeware clock generator tweaking utility ClockGen.

DFI also includes a system monitoring utility for the ITE IT8718F sensor.

DFI packs its Northbridge sink and audio module separately from the motherboard, but the rest of the installation kit looks rather basic. Users who prefer hardware overclocking will appreciate how DFI adds an extra package of jumpers with easy-grab tabs.

ECS Black Series X48T-A

Elitegroup Computer Systems has been at the high end a few times, formerly with its Extreme motherboard series and more recently with Nvidia reference designs. The company is stepping into a new direction in its “Black” series, shunning flashy board colors and noisy fans in an effort to provide enthusiasts what they really want : features, performance and stability.

Layout and Features

ECS really cleaned up the design of its X48T-A, but without removing many features. The most striking design elements include the enormous sinks of the heat pipe assembly, but the southward-positioned DIMM slots draw a second look.

The reason most companies are moving DIMM slots so far down the board is to shorten trace lengths for better memory stability at super-high speeds, but the trade off is loss of room for expansion cards. The X48T-A comes with only six slots, and the uppermost x1 slot is limited to cards no more than 7.1” (180mm) long. The X48T-A comes with only two PCI-Express x16 slots ; the top slot is in the third position on a case ; and both slots provide full PCI-Express 2.0 transfer modes.The elimination of a third x16 slot is no big loss, considering how competing brands are forced to provide theirs with only four lanes (at most) in PCI-Express 1.0 mode, but giving up a useless feature is still a gutsy move when so many buyers ignore practicality. Other expansion options include two PCI-Express x1 and two legacy PCI slots, but only one of each will be available in CrossFire configurations using double-slot cards.

But ECS’ design isn’t quite perfect, as two of its six internal SATA ports can easily be blocked off by the long coolers of double-GPU graphics cards. For buyers who don’t use such monster graphics, the benefit of facing the ports outward rather than forward is that they won’t be blocked by the hard drive cage of snug cases.

Like ASRock, ECS forces builders to stretch their front panel audio cables all the way back to the motherboard’s lower rear corner, which is a commonly used location that should have been done-away with long ago.

Maintaining a clean layout did require ECS to make a more serious compromise of placing the Ultra ATA connector under the lowest PCI slot, where its cable couldn’t possibly reach the upper drive bays of most cases, but the firm probably believes builders will opt for SATA optical drives instead. In the place the Ultra ATA connector “should have been,” we instead found a floppy cable header, which was ideally positioned at the front edge of the board, just above the center line. Many Windows XP users will be grateful that installing a floppy drive for RAID or AHCI drivers has been eased, and we think it’s ironic that the “presumed dead” floppy interface is now outliving Ultra-ATA thanks to Microsoft’s former OS installation requirements.

Moving back to the giant chunk of aluminum ECS uses to cool the Northbridge, we can see the heat pipes are glued in with thermal adhesive, rather than soldered on. This isn’t a problem as most companies would have you believe, as each of the sinks is generously sized for its specific task. In fact, ECS could have done away with heat pipes entirely, if not for the need to entice uninformed buyers.

Onboard Devices

What’s missing from the onboard device list ? High-end buyers have come to expect IEEE-1394 FireWire controllers, yet most don’t have any devices to connect to these. Designed for high-bandwidth peripherals, the most frequently found devices that still use these connectors are Sony video cameras. If you ever need one of these ports you’ll regret not having it, but most users never require them.

With FireWire not included, ECS instead provides a legacy serial port to fill the empty space on its port panel. We’d have preferred a second External SATA port (eSATA) instead, but ECS chose to use JMicron’s single-SATA plus Ultra ATA model JMB361 controller, rather than the more frequently picked JMB363.

One nice added feature of the X48T-A is its digital audio input, which we’ve not seen on many boards recently. The digital output is provided with both coaxial and optical connectors.

Five analog audio jacks support 5.1 surround plus microphone and line inputs, but 7.1 surround can be configured by disabling one of the inputs. This isn’t a significant loss, since the average user will employ the front panel header for their microphone.

Rated at a 97db signal-to-noise ratio, the ALC888S provides a total of 10 analog output channels. Users can output 7.1 surround signals to speakers while listening to a separate stream through their headphones.

ECS uses Intel’s high-end 82566DC GLCY PHY to access the ICH9R integrated gigabit networking capability, which is usually a smart move to conserve PCI-Express lanes. It appears, however, that the X48T-A actually has extra lanes left over.

The second Gigabit Ethernet port is provided by Realtek’s RTL8111B PCI-Express controller chip. Between this controller, the Ultra ATA/eSATA controller and the two x1 slots ; only four of the ICH9R’s six available PCI-Express lanes are used.

BIOS and Overclocking

The X48T-A by ECS offers an excellent range of FSB frequencies and adequate voltage adjustments for 65-nm and 45-nm cores, but no clock multiplier, FSB voltage or chipset voltage controls. Furthermore, its DRAM ratio adjustment appears to be broken on BIOS release 04/28/2008.

DRAM settings are available for all Intel ratios, but changing them doesn’t alter the actual value.

Without any FSB or Northbridge voltage control, we didn’t think we’d get very far in our overclock attempts. Later in this article our overclocking comparison will prove otherwise !

Software and Accessories

Our installation CD contained no value-added software or overclocking utilities, but instead just provided the drivers and some freeware.

The X48T-A accessory kit looks a little light, but does include all six Serial ATA cables, and all have 90 ends. Surprisingly, the company also included an eSATA cable.

Taking the “Black Series” name to the extreme, ECS left its company logo off the Ultra ATA cable.

Gigabyte GA-X48-DQ6

We broke our previous X48-DQ6 sample, so Gigabyte sent another. After many weeks of sitting on the shelf, we’ve finally found a good place to use it.

Our killing of the previous X48-DQ6 is far from unique to Gigabyte, as we’ve had motherboards from every brand fail. In fact, one of Gigabyte’s chief rivals is responsible for over half of our failed samples, and we’d probably get mixed up in a lawsuit if we named that rival here.

We’ve already covered the GA-X48-DQ6 design in full detail in our previous X48 Motherboard Comparison, so we’ll only cover the updates today.

And the change is…we went from BIOS F6 Beta (1/11/2008) to F6 Retail (2/29/2008). Even the motherboard PCB Revision 1.1 is the same.

Intel Extreme Series DX48BT2

Who would know a chipset better than the manufacturer? No longer simply a key player in OEM circles, Intel’s branded parts have been creeping into enthusiast markets for a few years. Will their better understanding of the underlying technology help them take the leadership role away from market partners?

Layout and Features

Anyone who didn’t think that Elitegroup’s black treatment went far enough should love Intel’s DX48BT2, with barely a scattering of white, blue and silver to lighten up the otherwise serious looking part.

Anyone who thinks that ASRock’s lack of heat pipes was a poor compromise towards lower cost might be shocked to see that Intel itself sees no need for such luxuries. Unlike lower-cost parts, however, Intel covers all of its CPU VRM MOSFET’s with sinks. Our apologies for the alphabet soup that made up the last comment.

The DX48BT2 provides the usual “two- -and-an-eighth” PCI Express 2.0 x16 slots, since the third slot is handicapped by both first-revision PCI-Express transfers and a maximum of four pathways. Unlike its partner brands, Intel has designed its board to assure that all four PCI-Express 1.0 pathways are available to the third slot at all times. That third slot might not be worthy of a gaming graphics card, but it’s certainly fast enough for most other duties.

Intel chipsets may no longer support Ultra ATA drives, but that didn’t keep the company from putting a connector on its board by way of a third-party controller. The Ultra ATA connector is ideally located in the top front corner, making it easy for cables to reach the top bays of even the tallest of “ordinary” cases. Likewise, ATX/EPS and ATX12V/EPS12V power connectors are within easy reach of top-mounted power supplies.

The DIMM slots are located relatively low on the DX48BT2, so removing memory with a graphics card installed is a little tedious. And if that graphics card happens to be super-sized, its cooler might also block the first Serial ATA port.

Intel has in fact moved the CPU, Northbridge and memory slots as a group away from the top edge of the board, which limits the number of expansion slots to six. There are no PCI-Express x1 slots, but the third x16 slot will support x1 cards nicely.

A power button is located on the DX48BT2’s bottom edge, making bench testing easy. Next to it is Intel’s traditional BIOS blocking jumper, which can be enabled from within BIOS to prevent average users from accessing BIOS without opening the case. To the left of the power button are Consumer Infrared Transmitter and Receiver device headers.

There is no floppy header in the DX48BT2, so Windows XP users who would like to install their OS on a RAID array must either slipstream the drivers onto a copy of their CD or use a USB floppy drive to install them at the F6 prompt. Intel is likely looking ahead into the not-to-distant future when XP is no longer available through standard channels.

Onboard Devices

Intel provides the full range of controllers we’d anticipate for any high-end board, including even the old IEEE-1394 FireWire interface, but the rear panel ports are a bit of a surprise:

In a not-so-bold move to eliminate legacy “junk,” Intel has eliminated PS/2 ports from the rear panel just as it removed the floppy cable header from inside. Two eSATA ports take their place, while a total of eight USB 2.0 ports provide more than enough connectivity for the average set of peripherals.

One other thing Intel has eliminated from its rear panel ports is any secondary Ethernet port. Intel is probably the biggest fan of Intel integrated controllers, and the chipset only directly supports one of those.

The 82566DC GLCI PHY accesses the ICH9R’s integrated networking capability, and being an Intel part we’d expect to find one on most of its desktop motherboards.

Intel gets its legacy Ultra ATA and modern eSATA connections from a single chip, the Marvell 88SE6121 PCI-Express to ATA controller.

From Ultra ATA, the next closest thing Intel has to legacy parts on its board is the Texas Instruments TSB43AB22A two-port IEEE-1394 controller.

Intel might be the “Leap Ahead” company, but it decided to “think different” about its audio solution. Formerly known as SigmaTel’s “Red River” audio controller, The IDT STAC9274D5 supports Dolby Home Theater for soundstage expansion (Dolby Headphone, Dolby Virtual Speaker) and on-the-fly analog to digital conversion (Dolby Digital Live), plus advanced features such as an integrated 12-band parametric equalizer and multi-streaming of separate 7.1 surround speaker and 2-channel headphone outputs.

BIOS and Overclocking

Intel has moved quite a distance from its previous position when it likened overclocking to thievery, as the company now provides a full set of adjustments for its high-flying DX48BT2 motherboard. Insane settings are available, such as FSB clocks up to 720MHz and a 1.600V maximum core voltage that can be extended to 1.900V by enabling the “CPU Voltage Offset” setting.

We found that the CPU Voltage Override (core voltage adjustment) did not work, however, as our Core 2 Duo E6850 retail CPU was always stuck at 1.35V. Enabling the CPU Voltage Offset did push it to 1.65V, but no software would detect the increase. We relied on a voltmeter to test these functions.

Intel even offers the GTL Reference voltage adjustment so highly sought after in top overclocking motherboard models.

All of the major memory timings are adjustable, as well as a few of the less familiar values. Intel doesn’t provide individual “automatic” modes for each setting, however.

Software and Accessories

Our installation kit included a large variety of documentation but just enough cables to install one SATA hard drive and two Ultra ATA optical drives! This “Reviewers Kit” might not represent everything included in retail boxes, however, as the company also left out the game “Ghost Recon 2” that the box says will be included with retail units.

Also included are a Northbridge fan bracket and a Southbridge badge. The bracket doesn’t include a fan, but one will be required in systems that lack adequate airflow around the Northbridge sink. The Southbridge badge will actually inhibit airflow, which is probably why Intel leaves it up to the builder to decide whether or not it should be installed.

Test Hardware

Once again we started with the CPU, RAM and graphics from our most recent reference system.

Two of today’s test platforms are DDR2-only, and one supports both DDR2 and DDR3. For DDR2 testing we used Crucial’s highly-overclockable Ballistix PC2-6400. In order to make this a fair fight between DDR2 and DDR3 motherboards, we set both the DDR2 and DDR3 memory to 1066 MHz data rate and CAS 5-5-5-15 timings during performance tests.

One motherboard that would not accept DDR2 at 1066 MHz data rate was the ASRock X48TurboTwins-WiFi. The setting simply wasn’t available. We tested it at DDR3-1066 for comparison to other platforms, plus DDR2-800 CAS 4-4-4-12 for comparison to itself.

In order to achieve our highest-possible “ambient air” overclock, we chose an enormous liquid cooling kit to keep our CPU near ambient temperatures. Swiftech’s Apogee GTX water block, MCP-655b high-flow pump and 120-mm x3 fan radiator have been keeping our systems on top for three consecutive LGA775 motherboard shootouts.

Weary from waiting hours for each hard drive controller test to complete, we sped things up with two Gigabyte I-RAM units, each filled with Crucial PC-3200 CAS 3.

Once again we had to make one exception in testing : While most motherboards supported RAID 0 through the third-party controller, the ECS X48T-A’s JMB361 controller supported only a single SATA drive, and was tested as such.

With a superb signal-to-noise ratio on its line input, the Asus Xonar D2X served as the input device for audio testing.

Because some testers don’t have access to some of the company’s software licenses, we ran today’s benchmarks on Windows Vista Ultimate.

Benchmark Settings

Benchmark Results: 3D Games

We don’t expect any large performance differences between motherboards that have the same chipset, but we still needed to prove that none of the boards had a performance issue. The only benchmark that stands out as even slightly unusual is Unreal Tournament 3, but the differences are small enough to ignore.

Applications

The Asus P5E3 Premium WiFi-AP @n Edition lagged significantly behind in our AVG virus scan, and we retested several times since that shouldn’t have happened. The ASRock X48TurboTwins-WiFi took both first and last place in DivX encoding, depending on whether it was using DDR3or DDR2, but that’s easily explained by how this particular motherboard doesn’t support DDR2 at a 1066-MHz data rate.

Synthetics

The DDR2 motherboards fall slightly behind in Sandra, even though we set the same timings and data rate.

Hard Drive Performance: H2benchw

We tested the third-party controller of each motherboard against the on-chipset RAID capability of the ICH9R, to see how each component performed. ASRock got left out, since it had no third-party SATA controller, while ECS was forced to run a single drive due to its single port.

Three of the motherboards tied for access times, while DFI’s implementation of the JMicron JMB363 somehow fell slightly behind.

Who would have thought that Marvell would take the lead away from JMicron in bandwidth ? Intel’s solution finishes far ahead.

Intel’s solution reads the fastest…

…and writes the fastest, too.

Hard Drive Performance: IOMeter

The only thing that affected the IOMeter File Server benchmark pattern is block size, where two of the controllers defaulted to 64k blocks and the other four defaulted to 128k blocks. Otherwise, the charts overlap. The single drive supported by ECS follows the higher path !

Audio Quality

We tested the onboard audio quality at 16 bits and 44.1 kHz to see how each codec handled this common standard.

The frequency response curve of the Intel solution appears to be altered by the built-in equalizer, even though we disabled advanced audio features for each test. All of the codecs drop off around 18kHz, but the DFI solution provides a slightly better range.

The noise level charts look messy, but every codec had a -110db noise level throughout most frequencies.

Dynamic Range looks good as well, but we don’t know what we could do with frequencies above 20000kHz if none of the codecs can play them at audible volume levels.

We see very little distortion in the sound signal of each onboard codec, except for an odd spike in the base range of Intel’s solution.

Power Consumption and Temperature

We tested power consumption and VRM temperature at default BIOS speed and voltage settings, with all power savings features enabled.

Asus has the most efficient board at idle, while several boards tied under full load. Gigabyte’s X48-DQ6 appears to fall in the middle, but that only happens when its Dynamic Energy Saver utility is manually enabled. Disabling DES caused the X48-DQ6 to fall into sixth place at 108W idle, 155W load. Furthermore, DES couldn’t be enabled when the system was overclocked.

Gigabyte does have a great VRM cooler, and we’re certain its 12-phase voltage regulator was also the least loaded. That puts the X48-DQ6 in the lead when it comes to temperature control.

Overclocking

Gigabyte might disagree with some competitors, but it owned them all in our FSB stability test.

DFI has put great effort into making its Lanparty series the best overclockers in the business, and the X48-T2R kept our E6850 stable at the highest clock speed. The lower-cost ECS X48T-A surprisingly comes out on top of the super-priced P5E3 Premium, in spite of the X48T-A’s lack of FSB and chipset voltage settings.

Asus continues its lead in memory stability, and the new motherboard revision is even better here than the last one was.

Performance Analysis

A brief recap of our performance findings should show more similarities than difference between each motherboard model. We’ll begin with games.

Asus wins the gaming performance roundup, but Intel’s latest Extreme Series motherboard isn’t far behind.

ECS takes the win in applications, with Asus dropping to the bottom.

Asus again loses in video encoding performance, but it’s a very tight race.

A board doesn’t have to win any particular category to come out on top overall, since two second-place finishes often beat a first- and last-place finish. Let’s see how these motherboards compare when we average all the performance differences.

Intel’s own-brand motherboard takes a surprise win in overall performance, but the P5E3 Premium isn’t far behind. ASRock spoils the curve by dragging it down with DDR2 installed, as a difference of less than 2% separates the rest of the field.

Conclusion

Once again we have no clear winner in a motherboard comparison, as most models were good in at least one area. Gigabyte had the highest FSB overclocking stability; DFI provided the best CPU overclocking; Asus supported the highest memory speeds; and Intel had the best overall performance. So which motherboard would we buy? It really depends on how we planned to use it.

Two models stood out, and both used DDR2 rather than DDR3 memory. DFI and Gigabyte had the best onboard feature set, but while Gigabyte also had the most complete set of rear panel ports, DFI had the highest stable DDR2 memory speed. The Lanparty LT X48-T2R also pushed our Core 2 Duo E6850 the farthest, but higher achievable FSB for the X48-DQ6 speed might put it ahead if we’d used a different processor.

If we had our hearts set on DDR3 memory, the battle would be between Asus and Intel. The DX48BT2 had the best overall performance, and placed second only to the DDR2-equiped DFI X48-T2R in maximum CPU clock. The P5E3 Premium WiFi-AP @n Edition supported higher memory and bus speeds, and the actual performance difference between the two was negligible. The DX48BT2 is a few features short of direct competition with the P5E3 Premium, but it’s also around $100 cheaper. How much are a few extra onboard components worth to you?

The Aurora m9700 sports Alienware's famous logo.

The Aurora m9700 sports Alienware’s famous logo.

The m9700 comes in "Cybord Green", "Saucer Silver" and "Conspiracy Blue". Green and Blue cost an additional $149. Most of the case has a matte finish. As you can see the keyboard deck has a mirrored finish.

The m9700 comes in "Cybord Green", "Saucer Silver" and "Conspiracy Blue". Green and Blue cost an additional $149. Most of the case has a matte finish. As you can see the keyboard deck has a mirrored finish.

Here's the DVD/CD drive, nicely located on the front of the notebook. The m9700 has a 17" 1920x1200 pixel display and is 15.6" wide, 11.75" deep and 2.00" high. The notebook weighs 9.4 pounds with its battery. The m9700 can also be purchased with a less e

Here’s the DVD/CD drive, nicely located on the front of the notebook. The m9700 has a 17" 1920x1200 pixel display and is 15.6" wide, 11.75" deep and 2.00" high. The notebook weighs 9.4 pounds with its battery. The m9700 can also be purchased with a less e

The Alienware logo is back-lit at the eyes and around the face.

The Alienware logo is back-lit at the eyes and around the face.

This notebook looks good from any angle.

This notebook looks good from any angle.

When the clamshell is closed you can see the logo, the cool black insets below the logo and how the top of the case peaks in the middle.

When the clamshell is closed you can see the logo, the cool black insets below the logo and how the top of the case peaks in the middle.

This is the front of the m9700 with the DVD/CD drive in the middle and speakers on the left and right ends.

This is the front of the m9700 with the DVD/CD drive in the middle and speakers on the left and right ends.

This is the bottom of the Aurora m9700. You access the most interesting parts inside the notebook through the bottom.

This is the bottom of the Aurora m9700. You access the most interesting parts inside the notebook through the bottom.

With the bottom part of the outer case removed you can see the graphics processor and CPU cooling assemblies mounted in the upper part of the case. The cooler at the top and to the left is for the Nvidia SLI graphics processors. The two X-shaped parts fit

With the bottom part of the outer case removed you can see the graphics processor and CPU cooling assemblies mounted in the upper part of the case. The cooler at the top and to the left is for the Nvidia SLI graphics processors. The two X-shaped parts fit

Here the cooler for the graphics processors has been removed and you can see the Nvidia GeForce Go 7900 GS graphics processors covered in thermal paste. We were unable to remove the cooler for the AMD Turion 2.4 GHz ML-44 Mobile CPU because one of the scr

Here the cooler for the graphics processors has been removed and you can see the Nvidia GeForce Go 7900 GS graphics processors covered in thermal paste. We were unable to remove the cooler for the AMD Turion 2.4 GHz ML-44 Mobile CPU because one of the scr

This is one of the Nvidia GeForce Go 7900 GS graphics processors. It and its mate come with 512 MB of dedicated video memory. You can also purchase a single 7900 GS or an SLI pair with 256 MB of video memory.

This is one of the Nvidia GeForce Go 7900 GS graphics processors. It and its mate come with 512 MB of dedicated video memory. You can also purchase a single 7900 GS or an SLI pair with 256 MB of video memory.

Markings indicate that this is indeed an SLI graphics notebook. As if the graphics processors themselves didn't scream SLI. :-)

Markings indicate that this is indeed an SLI graphics notebook. As if the graphics processors themselves didn’t scream SLI. :-)

The cooler for the graphics processors looks quite fierce when removed from the system.

The cooler for the graphics processors looks quite fierce when removed from the system.

This is one of the two 7200 RPM Seagate Momentus 100 GB hard disk drives in our Aurora m9700.

This is one of the two 7200 RPM Seagate Momentus 100 GB hard disk drives in our Aurora m9700.

The m9700's two hard drives were configured as RAID 1 in our test notebook. The notebook is also available in single drive and RAID 0 configurations.

The m9700’s two hard drives were configured as RAID 1 in our test notebook. The notebook is also available in single drive and RAID 0 configurations.

The notebook's battery specs at 14.8 volts and a fairly high 6450 milliamp hours, enough to keep the power hungry m9700 running somewhat longer than similar notebooks with more wimpy batteries.

The notebook’s battery specs at 14.8 volts and a fairly high 6450 milliamp hours, enough to keep the power hungry m9700 running somewhat longer than similar notebooks with more wimpy batteries.

The m9700's power supply weighs 2.32 pounds. So the notebook, its battery and power supply weigh almost 12 pounds (11.72 to be exact).

The m9700’s power supply weighs 2.32 pounds. So the notebook, its battery and power supply weigh almost 12 pounds (11.72 to be exact).

The back of the m9700 is so crammed with connectors that we'll need to show it to you in two segments. Click Next for the first segment.

The back of the m9700 is so crammed with connectors that we’ll need to show it to you in two segments. Click Next for the first segment.

This is the left side of the back of the Aurora m9700. From left to right are audio in, coaxial video in, S-Video in and the modem jack.

This is the left side of the back of the Aurora m9700. From left to right are audio in, coaxial video in, S-Video in and the modem jack.

More to the right on the back of the m9700 you'll find the power supply input, a USB 2.0 port, S-Video out, DVI in and VGA in.

More to the right on the back of the m9700 you’ll find the power supply input, a USB 2.0 port, S-Video out, DVI in and VGA in.

The left side of the Aurora m9700 contains so many connectors that we need to present them to you in three photos. Before we move on, however, note the two equal size slots on the bottom right of the case. The two hard disk drives slide into these slots.

The left side of the Aurora m9700 contains so many connectors that we need to present them to you in three photos. Before we move on, however, note the two equal size slots on the bottom right of the case. The two hard disk drives slide into these slots.

On the extreme left of the m9700's left side are the LAN jack, two USB 2.0 ports and an IEEE 1394 (Firewire) port.

On the extreme left of the m9700’s left side are the LAN jack, two USB 2.0 ports and an IEEE 1394 (Firewire) port.

Here's the memory card slot in the middle of the left side of the m9700. It supports SD, MS, MSPRO and MMC.

Here’s the memory card slot in the middle of the left side of the m9700. It supports SD, MS, MSPRO and MMC.

This is the PCI Express card slot on the extreme right of the left side of the m9700 case.

This is the PCI Express card slot on the extreme right of the left side of the m9700 case.

Here's the right side of the m9700. For a close-up of the jacks and ports on this side click Next.

Here’s the right side of the m9700. For a close-up of the jacks and ports on this side click Next.

From left to right are the volume control, headphone and microphone jacks, the center, surround and front left and right speaker outputs, an optical audio output and the fourth USB 2.0 port.

From left to right are the volume control, headphone and microphone jacks, the center, surround and front left and right speaker outputs, an optical audio output and the fourth USB 2.0 port.

Here is the Aurora m9700's keyboard. The notebook's larger keyboard and keys work quite well, with good tactile and audio feedback. Touch typing is easy once you learn the keyboard. The touchpad and button keys worked well after a bit of mouse driver adju

Here is the Aurora m9700’s keyboard. The notebook’s larger keyboard and keys work quite well, with good tactile and audio feedback. Touch typing is easy once you learn the keyboard. The touchpad and button keys worked well after a bit of mouse driver adju

Back left to right, top to bottom: LAN port, optical audio jack, air vent, power in, DVI out, 2 USB 2.0 ports (stacked), S-Video out, line out (headphones), line out, microphone in, IEEE 1394 Firewire port, two more USB 2.0 ports (stacked)

Back left to right, top to bottom : LAN port, optical audio jack, air vent, power in, DVI out, 2 USB 2.0 ports (stacked), S-Video out, line out (headphones), line out, microphone in, IEEE 1394 Firewire port, two more USB 2.0 ports (stacked)

Front left to right, top to bottom: power switch, CD/DVD drive slot, memory card reader (SD/MMC/MS/MS-pro), power light, USB 2.0 port, CD/DVD eject button

Front left to right, top to bottom : power switch, CD/DVD drive slot, memory card reader (SD/MMC/MS/MS-pro), power light, USB 2.0 port, CD/DVD eject button

Front and left side

Front and left side

Front and right side

Front and right side

Front and top

Front and top

The Shuttle XPC X100 operating vertically in its stand

The Shuttle XPC X100 operating vertically in its stand

AC power supply

AC power supply