"Wintel" alliance suffers as Intel cozies up to Chromebooks

New Chromebooks announced this week signal Intel's willingness to broaden its horizons and work with companies like Google, at the expense of its long-standing Windows partnership with Microsoft.

Three new Chromebooks from Hewlett-Packard, Acer and newcomer Toshiba with Google's Chrome OS were shown on stage during this week's Intel Developer Forum. The sub-$299 laptops will run on Intel's Haswell chips, and executives from Google and the chip maker said they worked closely to tune the OS at the kernel and driver levels to work with Intel's chips.

A tighter Chrome alliance with Google is just another example of how Intel, which has been largely left out of the tablet and smartphone markets, is spreading its wings to succeed in the PC, mobile and emerging markets like wearables. In an interview with IDG News Service, Intel president Renee James said the Microsoft-Intel alliance is alive, but the chip maker wants to offer choice beyond Windows.

"Microsoft [Windows] is not the only client operating system anymore. The same way for years and years Microsoft balanced between Intel and Advanced Micro Devices, we're in the same situation now. Our customers want choice, and we offer choice," James said.

"[Microsoft and Intel] get along in the face of the public, but there's huge animosity." 

Intel's partnership with Google is centered around Android, but the Chrome OS partnership is mutually beneficial. Intel is looking to perk up PC sales following the slow adoption of Windows 8, while Google is trying to expand beyond smartphones and tablets into desktop-style computing with Chrome OS.

"Chrome OS represents a new form of computing," said Sundar Pichai, senior vice president for Android, Chrome and applications at Google, in an appearance during a keynote speech by Doug Fisher, vice president of the software and services group at Intel, on Wednesday.

Intel sees Google as a partner for a range of devices, and the partnership has been getting stronger since they first started working together on Android two years ago, said Jack Gold, principal analyst at J. Gold Associates, during an interview on the show floor.

That doesn't mean the Wintel alliance isn't dead, but it is not close to what it was like four to five years ago, Gold said.

"Microsoft is going off dancing with other partners, Intel has got to do the same thing. Intel is going to go where the volumes are," Gold said, referring to Microsoft's decision to make Windows RT for processors designed by Intel's rival ARM Holdings.

GoogleA glimpse of the new Chrome devices announced Wednesday.

Intel had no choice but to support Chrome OS in the sub-$300 PC market, where Windows 8 does not fit, said Jim McGregor, principal analyst at Tirias Research.

"It's like netbook 2.0," McGregor said. "Intel is damned if they do, damned if they don't."

Intel's relationship with Microsoft is deteriorating, which also may have prompted the chip maker to approach Google, McGregor said. Intel tried to develop operating systems like Meego, but failed.

"[Microsoft and Intel] get along in the face of the public, but there's huge animosity," McGregor said.

The analysts agreed that Chrome OS won't sustain fast growth, but Intel had to continue supporting the OS to maintain a presence in the low-end of the PC market. Samsung offers a Chromebook based on a processor from ARM, which is trying to gain share in the low-cost PC market.

Intel will support any OS in markets ranging from PCs to wearables, but only if the customer asks for it, analysts said. For example, Intel has said smartphones based on its Atom chips could support Windows Phone OS, but only if customers requested it.

Intel this week introduced the very low power Quark family of chips for wearable and embedded devices. Intel didn't say which OSes will run on Quark, but Google has a non-Android OS for Google Glass. Intel also could possibly use a real-time operating system from its own Wind River unit. But the chip maker needs to keep its options open, analysts said.

"It's a logical business model for Intel," Gold said.

Agam Shah is a reporter for the IDG News Service in New York. He covers hardware including PCs, servers, tablets, chips, semiconductors, consumer electronics and peripherals.
More by Agam Shah, IDG News Service

View the original article here

IDF 2013 keynote: Intel announces Bay Trail tablet CPU, slew of Haswell Chromebooks

Intel on Wednesday ended weeks of speculation by launching "Bay Trail," the next-generation Atom Z3000 chip that pushes tablets towards PC-esque levels of performance.

A few years ago, Intel launched the Atom processor, designed for an emerging low-power world.  But over time, Atom's performance has improved to the point that the most recent iteration, Bay Trail, has improved dramatically in PCWorld's hands-on tests.

On Wednesday, Intel touted Bay Trail, which looks like it will cut into Core-based PCs at the low end of the desktop and notebook, and the convertible tablets or "two-in-one" products that will use them. Intel also announced that OEMs would be including a Kinect-like sensor into future PCs, as previously reported, before touching on a wave of next-gen, Haswell-powered Chromebooks slated to land in the coming months.

Hermann Eul, vice president and general manager of the Mobile Communications Group, described how Intel is designing multiple components across the system. "[Mobility] connects us to our partners, to our friends, and to the cloud—even to our own bodies," Eul said.

Intel described its "soup to nuts" approach to mobility: a CPU, an imaging processor, as well as graphics connected to it all. That device, in turn, will be connected via short-range and long-range WWAN technologies, with security layered on top. And on that, Intel can layer its own software, with dozens of investments that Intel has made in the embedded space.

"This, we call a platform: from A to Z," Eul said. "And it all starts with a great CPU."

That core is the "Silvermont," the design revision that forms the underlying technology of the Bay Trail and other chips within Intel. Silvermont is a 64-bit architecture, which Intel revealed a few months ago.

"Bay Trail is architected for the best mobile computing experience," Eul said. It provides leading battery life as well as ample performance on demand, he said, plus a next-generation imaging core.

Intel executives showed off a Bay Trail miniature tablet running and editing video and invited Jerry Shen, chief executive of Asus to introduce the T100, a Bay Trail notebook with over ten hours of battery life. "We are very excited about the Bay Trail quad-core promise," he said.

Dell was also invited on stage. Neil Hand, vice president of tablets, introduced an eight-inch Bay Trail tablet that will "innovate and drive new capabilities," he said. Dell will be branding the new family of tablets with the "Venue" brand, unveiling the entire range of products on Oct. 2 in New York City.

Eul also characterized Bay Trail as a "wonderful gaming platform," optimized for both Windows and Android. Intel also brought up Victoria Molina, a former executive for Ralph Lauren, Levi's, and the Gap, to show off a Bay Trail-powered virtual shopping experience, where a shopper "tried on clothes" with a model who had uploaded her measurements.

Merrifield represents the Silvermont architecture in the phone, with all the performance, battery life, and security components of the Bay Trail platform. Advanced LTE is coming in 2014. Airmont will follow with 14-nm technology for the phone, Eul said.

Kirk Skaugen, senior vice president and general manager of the PC Client Group, then turned to two-in-one devices. In 2011, Intel launched the Ultrabook. "Now we've stopped counting [OEM designs], and assumed that the entire world has gone thin," Skaugen said.

More than 40 percent of all Core notebooks have been designed with touch. Seventy percenty of today's Ultrabooks are touch-enabled, on the way to 100 percent touch later this year, creating a "bow wave" driving touch PCs, Skaugen said.

Skaugen described the two-in-one as the "best of a laptop and the best of a tablet," with a 10-inch screen size, a "full PC OS" like Windows 8, and an integrated keyboard design.

More than 80 percent of tablet users say they want to refresh their notebook, Skaugen said, and the company said it believes it can capture a portion of the tablet market. Over 60 two-in-one designs will be in the marketplace by year's end, he said.

A on-stage display showed off the Sony Duo 13-inch slider, the Dell XPS 11, and a Sony two-in-one. An application from CyberLink will be provided on Haswell machines by the end of the year to facilitate content creation, presumably video editing.

Intel also invited Microsoft on stage. Tami Reller, the executive vice president for Microsoft in charge of Windows, said that Microsoft were "fans of the category".

More than 2 million people have downloaded Windows 8.1, Reller said, although she didn't say if that was simply those who downloaded the preview, or the final version via MSDN or TechNet. Microsoft saw the highest number of Windows 8 activations in August, a new record. (Read PCWorld's review of Windows 8.1.) A Facebook app is coming, Reller reiterated, while Foursquare recently released its own Windows 8 app.

GOOGLEThe next generation of Chromebooks.

Finally, Intel also said that a new, unannounced generation of Chromebooks would use the existing "Haswell" Core processor inside Chromebook and Chromeboxes from Acer, Asus, HP, Toshiba, and others. Toshiba and Asus are new to the Chrome family; Google says Asus will supply a Chromebox, rather than a standalone Chromebook.

"With today’s announcement, now six of the top laptop manufacturers are offering Chromebooks," Google's announcement bragged.

View the original article here

Intel hires wearable computing talent, but it's late to the game

Intel said Friday that it has hired a pair of high-level engineers from Nike and Oakley to assist the company with building wearable computing devices.

Intel said that it had hired Hans Moritz, described as the man who led development of the Oakley AirWave heads-up goggles, to work at the “new devices” team being formed within Intel. Moritz has spent his entire 16-year career at Oakley, where he most recently led the integration of the Oakley Airwave smart ski goggles pictured above.

At Intel, Moritz will join Steven Holmes, most recently of Nike, where he was responsible for the end-to-end development of the Nike GPS Sportswatch as well as the Fuelband fitness tracker accessory. Holmes was hired by Intel a year ago. The Moritz hire was reported by the Intel Free Press, described as a news organization within Intel run by former members of its communications team.

Both will work for Michael Bell, a former vice president at Apple and former senior vice president for product development at Palm, who leads the “new devices” outfit. So far, Intel has said only that it plans a Roku-like set-top box to stream video to TVs.

At the time of his hire in 2010, Bell’s mission was to “lead a team with the charter to build breakthrough smartphone reference designs with the explicit intent of accelerating Intel Architecture into the market.” Since then, however, his role has apparently broadened into the taking Intel’s chips into new product areas, regardless of which products they fall into.

Intel has struggled somewhat in the embedded and mobile space, as low-power RISC designs from ARM and other embedded architectures have moved into phones and other devices. To answer, Intel has developed a series of low-power Atom processors to bring full X86 compatibility to the embedded market. Next week, at the Intel Developer Forum, the company has scheduled press briefings on Bay Trail, a next-generation Atom processor designed for phones, tablets, and other mobile devices.

So far, however, Intel has said nothing about whether or not the Atom can play inside smartwatches or eyewear like Google Glass, which have even more demanding space and power requirements than even miniature tablets. Intel, for its part, has been a heavy promoter of failed devices like Mobile Internet Devices, and, most recently, ultrabooks. Intel’s existing Clover Trail+ processor for phones has scored design wins in Asia, but not within the United States—in part because of a lack of integrated LTE connectivity.

“Intel has a history of trying to dictate to the market; we just enable people to do stuff,” said James Bruce, a director of mobile strategy for ARM, in an interview this week.

It’s not clear whether or not Intel will adopt the same strategy in wearable computing. Moritz didn’t design the heads-up display within the AirWave; that was done by Recon Instruments, which has its own line of HUD modules that project information onto glasses through a series of connected microprojectors. But the knowledge of Holmes and Moritz will at point Intel in the right direction. Whether it or not can catch up is another question.

View the original article here

From calculators to tablets, Intel focuses on the mobile era

Intel’s dominance of the chip market is starting to wane as PC shipments slump and smartphone and tablet adoption grows, but the manufacturer will try to prove it can make fast and power-efficient processors for mobile devices at its annual developer gathering next week.

At the Intel Developer Forum next week, the company will introduce low-power Atom chips code-named Bay Trail, which will go into tablets that are priced from $150. The show will also be a litmus test for new CEO Brian Krzanich, who will deliver the keynote speech on Sept. 10 and highlight Intel’s long-term strategy as the company diversifies into the mobile market.

PCWorld (US)Intel CEO Brian Krzanich

Krzanich replaced the popular Paul Otellini in May this year, and has put mobile chips at the top of his priority list. He has also created a “new devices” group that will focus on emerging areas like wearables. The chip maker is already planning to release a TV service through its own set-top box, and Krzanich may touch upon the wearable devices during his opening keynote on Sept. 10.

A distraction will be an Apple product launch event on Sept. 10, which starts just one hour after Krzanich’s keynote begins.

Apple’s iPhone 5 launch event was held last year on Sept. 10, one day ahead of IDF, but chatter about the smartphone and other Apple products lingered on the IDF floor for the entire show. Apple is expected to launch new iPhones at its event.

The transition to mobile is one of the more challenging issues for Intel since the chip maker made its first 4004 chip for calculators in 1971. Aided by Windows, Intel went on to thrive in the PC market, but is now just catching up to ARM in tablets and smartphones. Though the Wintel alliance remains alive, Intel has now adopted the widely used Android operating systems for mobile devices as it tries to gain market share.

Intel has also struggled to dispel a notion that its low-power mobile chips are watered-down versions of its power-hungry x86 PC chips, an idea the company will try to debunk at the show through its Bay Trail chips. Based on the Silvermont architecture, the chips have an all-new design and will be available in dual-core and quad-core variants.

Intel claims Bay Trail tablets will have longer battery life and better performance compared to its Atom chips. Intel hopes Bay Trail will be a legitimate competitor to ARM, and the companies have already traded barbs on superiority in performance and power efficiency.

Intel has said Bay Trail tablets will start at $150. The first few models will likely be priced at about $400 or more and run Windows 8.1, but some Android models shipping toward the end of the year will be lower in price.

Tablets with ARM processors and Android are available for under $100 in some cases, though some popular brands such as Google’s Nexus 7 start at over $200.

Intel is also extending Bay Trail chips to low-cost Windows 8, Android and Chrome OS laptops and desktops. The company will also highlight developments in Ultrabooks, which are thin and light laptops that can also function as tablets.

Adoption of Ultrabooks has been poor due to high product prices, but Lenovo and Dell in the past month have announced new laptops and hybrids based on Intel’s fourth-generation Core processors, code-named Haswell.

The chip maker is also expected to announce new server chips, and will also touch upon new data center technologies such as optical networking. Intel is tuning many of its server chips specific to software applications, and many technology briefing sessions will be focused on software development for big data, supercomputing and mobile devices.

Intel is also dabbling in the area of open-source PCs and the “Internet of things.” In addition, the company will demonstrate Thunderbolt 2, which will transfer data between host PCs and peripherals at twice the speed of its current 10G bps (bits per second) rate.

Agam Shah is a reporter for the IDG News Service in New York. He covers hardware including PCs, servers, tablets, chips, semiconductors, consumer electronics and peripherals.
More by Agam Shah, IDG News Service

View the original article here

Intel pushes speed, reliability claims with its new MXC cable

IDG News Service - An optical interconnect introduced by Intel on Wednesday may someday slim down cabling throughout data centers if the company can get enough vendors to mass-produce it.

The interconnect, which Intel is calling MXC, is designed to offer high speed with a long reach and a relatively low cost. Intel developed MXC with Corning and showed it off at a press event in San Francisco. Its next goal is to publish the MXC specification so fiber manufacturers can start using it to develop new products, said Victor Krutul, director of business development and marketing at Intel's Silicon Photonics Operation.

MXC is protocol-agnostic and could be used for network links throughout a data center, over technologies that could include Terabit Ethernet, Krutul said. At Wednesday's event, Intel used it to connect two trays of microservers to each other in a demonstration rack. The fiber could also be used to link servers to a top-of-rack switch, connect a series of those switches to a bigger switch at the end of a row, or form the backbone of the data-center network, Krutul said.

A key application where MXC may be needed is so-called rack-level computing, where the various components found in servers are separated and packed in large numbers into rack units of their own, Krutul said. Rack-level computing needs to bring together parts distributed throughout a data center, an ideal fit for MXC's range and speed, he said.

Intel gave a few details about MXC in its description of a session at next week's Intel Developer Forum. The announcement on Wednesday came alongside other signs that the company is making another push into networking, a market where it has had mixed results. Intel also announced the Atom C2000 low-power processor line, which will come in system-on-a-chip variants for switches and routers, among other uses.

MXC can carry 25Gbps (bits per second) on each fiber over a distance of 300 meters, a combination of speed and range that today's most comparable fiber technology, VCSEL (vertical cavity surface-emitting laser), can't achieve, said Ken Chong, an Intel silicon photonics product line manager. VCSEL offers 10Gbps over 300 meters now, but at 25Gbps it can only go 100 meters or less, he said.

As many as 64 fibers can be bundled into one MXC cable, bringing its total capacity to as much as 1.6Tbps, Krutul said. That means MXC can be used for the 10Gbps connections commonly used in data centers today as well as newer 40Gbps and 100Gbps versions, up to the Terabit Ethernet that Ethernet researchers envision in the future, he said. Other protocols, such as PCI Express, could also ride over MXC fiber. MXC components convert the electronic signals at each end of a connection into optical signals, then back again at the other end.

Reprinted with permission from IDG.net. Story copyright 2012 International Data Group. All rights reserved.

View the original article here

Intel pushes speed, reliability claims with its new MXC cable

IDG News Service - An optical interconnect introduced by Intel on Wednesday may someday slim down cabling throughout data centers if the company can get enough vendors to mass-produce it.

The interconnect, which Intel is calling MXC, is designed to offer high speed with a long reach and a relatively low cost. Intel developed MXC with Corning and showed it off at a press event in San Francisco. Its next goal is to publish the MXC specification so fiber manufacturers can start using it to develop new products, said Victor Krutul, director of business development and marketing at Intel's Silicon Photonics Operation.

MXC is protocol-agnostic and could be used for network links throughout a data center, over technologies that could include Terabit Ethernet, Krutul said. At Wednesday's event, Intel used it to connect two trays of microservers to each other in a demonstration rack. The fiber could also be used to link servers to a top-of-rack switch, connect a series of those switches to a bigger switch at the end of a row, or form the backbone of the data-center network, Krutul said.

A key application where MXC may be needed is so-called rack-level computing, where the various components found in servers are separated and packed in large numbers into rack units of their own, Krutul said. Rack-level computing needs to bring together parts distributed throughout a data center, an ideal fit for MXC's range and speed, he said.

Intel gave a few details about MXC in its description of a session at next week's Intel Developer Forum. The announcement on Wednesday came alongside other signs that the company is making another push into networking, a market where it has had mixed results. Intel also announced the Atom C2000 low-power processor line, which will come in system-on-a-chip variants for switches and routers, among other uses.

MXC can carry 25Gbps (bits per second) on each fiber over a distance of 300 meters, a combination of speed and range that today's most comparable fiber technology, VCSEL (vertical cavity surface-emitting laser), can't achieve, said Ken Chong, an Intel silicon photonics product line manager. VCSEL offers 10Gbps over 300 meters now, but at 25Gbps it can only go 100 meters or less, he said.

As many as 64 fibers can be bundled into one MXC cable, bringing its total capacity to as much as 1.6Tbps, Krutul said. That means MXC can be used for the 10Gbps connections commonly used in data centers today as well as newer 40Gbps and 100Gbps versions, up to the Terabit Ethernet that Ethernet researchers envision in the future, he said. Other protocols, such as PCI Express, could also ride over MXC fiber. MXC components convert the electronic signals at each end of a connection into optical signals, then back again at the other end.

Reprinted with permission from IDG.net. Story copyright 2012 International Data Group. All rights reserved.

View the original article here

Intel pushes speed, reliability claims with its new MXC cable

IDG News Service - An optical interconnect introduced by Intel on Wednesday may someday slim down cabling throughout data centers if the company can get enough vendors to mass-produce it.

The interconnect, which Intel is calling MXC, is designed to offer high speed with a long reach and a relatively low cost. Intel developed MXC with Corning and showed it off at a press event in San Francisco. Its next goal is to publish the MXC specification so fiber manufacturers can start using it to develop new products, said Victor Krutul, director of business development and marketing at Intel's Silicon Photonics Operation.

MXC is protocol-agnostic and could be used for network links throughout a data center, over technologies that could include Terabit Ethernet, Krutul said. At Wednesday's event, Intel used it to connect two trays of microservers to each other in a demonstration rack. The fiber could also be used to link servers to a top-of-rack switch, connect a series of those switches to a bigger switch at the end of a row, or form the backbone of the data-center network, Krutul said.

A key application where MXC may be needed is so-called rack-level computing, where the various components found in servers are separated and packed in large numbers into rack units of their own, Krutul said. Rack-level computing needs to bring together parts distributed throughout a data center, an ideal fit for MXC's range and speed, he said.

Intel gave a few details about MXC in its description of a session at next week's Intel Developer Forum. The announcement on Wednesday came alongside other signs that the company is making another push into networking, a market where it has had mixed results. Intel also announced the Atom C2000 low-power processor line, which will come in system-on-a-chip variants for switches and routers, among other uses.

MXC can carry 25Gbps (bits per second) on each fiber over a distance of 300 meters, a combination of speed and range that today's most comparable fiber technology, VCSEL (vertical cavity surface-emitting laser), can't achieve, said Ken Chong, an Intel silicon photonics product line manager. VCSEL offers 10Gbps over 300 meters now, but at 25Gbps it can only go 100 meters or less, he said.

As many as 64 fibers can be bundled into one MXC cable, bringing its total capacity to as much as 1.6Tbps, Krutul said. That means MXC can be used for the 10Gbps connections commonly used in data centers today as well as newer 40Gbps and 100Gbps versions, up to the Terabit Ethernet that Ethernet researchers envision in the future, he said. Other protocols, such as PCI Express, could also ride over MXC fiber. MXC components convert the electronic signals at each end of a connection into optical signals, then back again at the other end.

Reprinted with permission from IDG.net. Story copyright 2012 International Data Group. All rights reserved.

View the original article here

Intel pushes speed, reliability claims with unveiling of MXC cable

An optical interconnect introduced by Intel on Wednesday may someday slim down cabling throughout data centers if the company can get enough vendors to mass-produce it.

The interconnect, which Intel is calling MXC, is designed to offer high speed with a long reach and a relatively low cost. Intel developed MXC with Corning and showed it off at a press event in San Francisco. Its next goal is to publish the MXC specification so fiber manufacturers can start using it to develop new products, said Victor Krutul, director of business development and marketing at Intel’s Silicon Photonics Operation.

MXC is protocol-agnostic and could be used for network links throughout a data center, over technologies that could include Terabit ethernet, Krutul said. At Wednesday’s event, Intel used it to connect two trays of microservers to each other in a demonstration rack. The fiber could also be used to link servers to a top-of-rack switch, connect a series of those switches to a bigger switch at the end of a row, or form the backbone of the data-center network, Krutul said.

A key application where MXC may be needed is so-called rack-level computing, where the various components found in servers are separated and packed in large numbers into rack units of their own, Krutul said. Rack-level computing needs to bring together parts distributed throughout a data center, an ideal fit for MXC’s range and speed, he said.

Intel gave a few details about MXC in its description of a session at next week’s Intel Developer Forum. The announcement on Wednesday came alongside other signs that the company is making another push into networking, a market where it has had mixed results. Intel also announced the Atom C2000 low-power processor line, which will come in system-on-a-chip variants for switches and routers, among other uses.

MXC can carry 25Gbps (bits per second) on each fiber over a distance of 300 meters, a combination of speed and range that today’s most comparable fiber technology, VCSEL (vertical cavity surface-emitting laser), can’t achieve, said Ken Chong, an Intel silicon photonics product line manager. VCSEL offers 10Gbps over 300 meters now, but at 25Gbps it can only go 100 meters or less, he said.

As many as 64 fibers can be bundled into one MXC cable, bringing its total capacity to as much as 1.6Tbps, Krutul said. That means MXC can be used for the 10Gbps connections commonly used in data centers today as well as newer 40Gbps and 100Gbps versions, up to the Terabit ethernet that ethernet researchers envision in the future, he said. Other protocols, such as PCI Express, could also ride over MXC fiber. MXC components convert the electronic signals at each end of a connection into optical signals, then back again at the other end.

There are longer-range fiber systems, but they use single-mode fiber, which is more expensive and harder to install than multimode fiber, which MXC uses. Cables made with this type of fiber are highly flexible and much thinner than typical copper cable. That difference can add up in a data center where hundreds of thousands of servers need to be linked together, said Chris Phillips, general manager of Windows Server and System Center program management at Microsoft. It’s one of the things that has to change in data centers, he said.

“Wires really are hard,” Phillips said.

Masses of cables can restrict air flow and get in the way of technicians working in the data center, making it more likely they’ll knock one out of a port, Phillips said.

MXC is designed to prevent that type of accident, as well as breaks due to dust, the biggest threat to fiber in data centers, Intel’s Kruzul said. The connectors Intel has developed for MXC can withstand up to 45 pounds of pull force before getting accidentally dislodged, compared with 10 or 11 pounds for current fiber connectors, Kruzul said. In addition, a “beam expander” feature built into the connector spreads out the highly focused beam of light so a piece of dust won’t block the entire beam, he said.

The connector also is designed for low cost and greater reliability with just seven parts, compared with 27 parts for existing fiber connectors, he said.

Intel wants to publish the MXC specification so third-party cabling vendors can make products for it, driving volume up and costs down, Kruzul said. But the company has to work through some issues, such as intellectual property, before going ahead, he said. It may also seek to have MXC standardized in cable organizations and the Institute of Electrical and Electronics Engineers (IEEE), but those processes tend to take a long time, Kruzul said.

View the original article here

Intel Core i7-4960X

Pros The fastest consumer processor money can buy for multi-threaded workloads. Asus makes a great motherboard now that Intel has killed its own manufacturing division.

Cons Lacks support for latest Haswell features. Not the fastest chip for single-threaded tasks. Not compatible with Intel-manufactured motherboards. Bottom Line The Intel Core i7-4960X processor updates the CPU architecture to the 22nm-based Ivy Bridge, but makes precious few additional changes.

By Joel Hruska

It has been two years since Intel launched its six-core Sandy Bridge-E series and its accompanying LGA2011 socket. In 2011, Sandy Bridge-E was Intel's first consumer-oriented product with quad-channel DDR3-1600 memory, and support for PCI-Express 3.0 (though not on Intel's own motherboards). The Intel Core i7-4960X processor launching today updates the CPU architecture to the 22nm-based Ivy Bridge, but makes precious few additional changes.

The new Ivy Bridge-E CPUs will be available in three variants. At the top-end part is the hexa-core Core i7-4960X, with a clock speed of 3.6GHz base and 4GHz Turbo. That's a modest bump over the Intel Core i7-3960X, with its base clock of 3.3GHz and a 3.9GHz Turbo. The L3 cache is 15MB on both models, but the new i7-4960X does support slightly faster memory—quad-channel DDR3-1866 is formally supported, up from the Sandy Bridge-E's DDR3-1600. The Core i7-4960X is set to debut at $990, which matches the launch price of the Intel i7-3960X.

There will be two additional IVB-E processors—the Intel Core i7 4930K, at 3.4GHz Base/3.9GHz Turbo and $555, and the quad-core Intel Core i7 4820K, at 3.7GHz Base/3.9GHz Turbo, with 10MB of L3 cache and a list price of $310. That puts the 4820K just slightly below the Intel Core i7-4770K, but in this case, that distinction is a tad misleading. The 4770K is based on Haswell, Intel's newer architecture, and is 5 to 8% faster than Ivy Bridge when measured clock for clock. The Intel Core i7-4820K, in other words, is almost certainly going to be outperformed by the Haswell processor.

Since Sandy Bridge and Ivy Bridge performance were very nearly identical, there's not a lot to get excited about here. Multi-threading performance is going to improve modestly, thanks to an increase in non-Turbo base clock speed. Single-threaded performance between the older Intel Core i7-3960X and newer Core i7-4960X is going to be nearly identical, which means we may see the Haswell-based 4770K outperforming both cores in some tests.

Backwards-Compatiblity
One of the benefits to a relatively small update like this is that backwards compatibility is much easier to ensure. Sure enough, if you own an X79 motherboard, you can look forward to an Ivy Bridge-E upgrade, however marginal that upgrade might happen to be—unless you own Intel's X79 motherboard. If you had the misfortune to purchase an Intel DX79SI, DX79SR or DX79TO (boards that retailed for between $219 and $319), you don't get to upgrade.

When Intel announced it would kill off its motherboard division earlier this year, it was understood that this meant the company wouldn't be building new motherboards for its own chipsets. The company neglected to mention that this also meant it would immediately dump all support for its enthusiast products, including refusing to provide a minor microcode BIOS update. Typically, adding support for a new processor is a relatively simple affair, particularly when the new chip doesn't introduce new features, core counts, instruction sets, or operating parameters.

It's worse, in this case, because Intel had a two-decade reputation for building solid motherboards with excellent customer support. Users who bought on the strength of that reputation two years ago, however, are out of luck here.

Test Results
We tested the 4960X using an AMD Radeon 7990 dual GPU, 16GB of Mushkin DDR3-1866, and an Asus X79-Deluxe motherboard. The Asus board is a particularly gorgeous implementation of the chipset, with six USB 3.0 ports in back, dual gigabit Ethernet, integrated Wi-Fi, and support for up to 64GB of RAM across 8 DIMMs. Intel's specifications state that the system should only use one DIMM per channel in a quad-channel configuration at this speed, however, so that's what we've gone with. All tests were run using a SanDisk Extreme II 240GB SSD for primary storage, with Windows 7 64-bit SP1 (all additional patches and updates were installed prior to testing.)

Is the 4960X faster than the Intel 3960X or the Haswell-based 4770K? Barely, compared with the first, and sometimes, when it comes to the second. It's important to pay attention to where the chip's benefits do and don't materialize. In 3D rendering test Cinebench 11.5, the 3960X scored a 1.57 in the single-threaded test, compared to the 4960X's 1.67. That's a gain of 4% for the Ivy Bridge-E chip, but the Core i7-4770K scores a 1.78, putting it a further 9% ahead. In the multithreaded test, where the hexa-threaded processors can stretch their legs, the Core i7-4960X scored an 11.13 compared with 10.52 for the Core i7-3960X and 8.18 for the Core i7-4770K. That's a gain of 6% compared to the Sandy Bridge-based processor and 36% over the quad-core Haswell.

In the encryption and security benchmark test TrueCrypt 7.1a, both of the six-core processors beat out the Intel Core i7-4770K. The Core i7-4960X and Core i7-3960X both scored 297MBps, while the 4770K trailed the pack, at 240MBps in the Serpent-TwoFish-AES test. In Handbrake, in our iPod Touch file conversion test, the Core i7-4960X again won out overall, with a 25-second convert time compared to 28 seconds for the Intel Core i7-3960X and 29 seconds for the Intel Core i7-4770K.

Where the 4960X runs into trouble is in benchmark tests that don't give it room to stretch its cores. In POV-Ray 3.6, the 4960X took 247 seconds to render the benchmark scene, compared to 256 seconds for the Core i7-3960X took 256 seconds. The Haswell-based Core i7-4770K finished the job in 193 seconds. In Adobe Photoshop CS6, using our 11-filter test, the Core i7-4960X finished in 189 seconds, compared with 169 seconds for the Core i7-4770K.

In the general-purpose PCMark 7 test, the Haswell-based Core i7-4770K actually pulled ahead, with a score of 6,686, compared with 5,917 for the Core i7-4960X. Obviously, PCMark 7 isn't optimized for more than four cores, but that's indicative of the general problem—a great deal of modern software isn't really designed to run at up to 12 threads. Most games, for example, won't really benefit.

Split Verdict
The value of the Intel Core i7-4960X is much weaker than it had it to be, thanks to multiple compromises on core counts and small clock speed increases. Intel is poised to launch a number of eight-core Xeon processors— moving one of those chips over to LGA2011 while refreshing the chipset to add native USB 3.0 support and more SATA 6G ports could have been a reasonable update for the platform.

When Sandy Bridge-E launched, the LGA2011 platform and six-core processor were the fastest Intel solutions you could buy. Two years later, the Ivy Bridge-E is the fastest solution, but only some of the time. For a chip that costs three times as much as the Intel Core i7-4770K, but lacks certain features and capabilities, that's a poor value.

Here's the bottom line: If you have multi-threaded workloads that you know respond well to more than four cores, then yes, the Intel Core i7-4960X is still going to be an interesting chip but only if you buy the highest end. Unlike in 2011, where the lower-end Intel Core i7-3820X might have been a valid option to pair with the X79 chipset for PCI-Express 3.0 and more memory bandwidth, there's no reason to pay a premium for a last-generation X79-compatible processor like the i7-4960X now as compared to a Core i7-4770K. In a core-for-core comparison, Haswell is always going to beat Ivy Bridge.

View the original article here

Intel wants to help gamers, others overclock their SSDs

Computerworld - Without saying exactly what it will be, Intel plans to make a "big splash" near the end of the year or the beginning of next with a product that can "overclock" solid-state drives (SSDs).

"This is a product we're looking into but we have not released yet. So I'm not able to go into great detail about a future product ... as far as specs and such," said Alan Frost, marketing and communications manager for Intel.

What Frost could say is that the SSD product will allow users to tweak the percentage of the drive that's used for data compression.

Overclocking typically refers to pushing past the recommended processor clockspeed, thereby increasing performance over what is specified by a manufacturer. In the case of SSDs, Intel is using the term "overclocking" to define modifying its specified SSD parameters.

At its Intel Developers Forum next month in San Francisco, Intel has scheduled an information session on overclocking SSDs. Frost said the session will remain at a high level, and it will not reveal specifics about the product.

The conference session, titled "Overclocking Unlocked Intel Core Processors for High Performance Gaming and Content Creation," is aimed at system manufacturers and developers as well as do-it-yourself enthusiasts, such as gamers.

"Anyone interested in the performance tuning and overclocking experience on desktop and mobile platforms," the session description states.

Frost stressed that Intel executives are still talking in back rooms about what markets might benefit from creating an SSD with flexible provisioning properties.

"We've debated how people would use it. I think the cool factor is somewhat high on this, but we don't see it changing the macro-level environment. But, as far as being a trendsetter, it has potential," Frost said.

Michael Yang, a principal analyst with IHS Research, said the product Intel plans to release could be the next evolution of SandForce controller, "user definable and [with the] ability to allocate specified size on the SSD."

"Interesting, but we will have to see how much performance and capacity [it has] over existing solutions," Yang said in an email reply to Computerworld.

Late last year, Intel began using the third-party SandForce (now owned by LSI) controller for the first time in its SSD 520 flash drive.

"The premise of SandForce SSD controllers is compression. By compressing data, less time is required to transfer data from and to the SSD," Yang wrote. "Is Intel talking about its own controller? Or are they simply 'glamorizing' something they already have?"

Data compression would first and foremost increase the capacity of a drive, but it could also lead to greater performance. Frost cautioned that allocating a percentage of an SSD's capacity for compression would come at a cost. That cost could be shorter endurance, according to analysts.

"If you overclock and get faster performance and capacity and sacrifice endurance by doing so, well, then you could still enjoy the benefits if you are in a mostly read-intensive environment," said Joseph Unsworth, Gartner's NAND flash and SSD research vice president. "I suspect DataCenter server IT admins would also be savvy enough (especially at hyperscale level) to exploit this advantage."

Frost made it clear that the SSD overclocking product won't be for the average consumer, whom he still thinks is learning what the term "solid-state drive" means.

"SSD penetration is getting better, but still small overall," Frost said. "The market for this is not big."

This article, Intel wants to help gamers, others overlock their SSDs, was originally published at Computerworld.com.

Lucas Mearian covers storage, disaster recovery and business continuity, financial services infrastructure and health care IT for Computerworld. Follow Lucas on Twitter at @lucasmearian or subscribe to Lucas's RSS feed . His e-mail address is lmearian@computerworld.com.

See more by Lucas Mearian on Computerworld.com.

Read more about SSD in Computerworld's SSD Topic Center.

Small and midsize businesses are moving to the cloud to host their communications capabilities. Learn how enterprise-quality phone benefits, online management, conferencing, auto attendant, and ease of use are built into a system that is half the cost of a PBX.

Read now.

View the original article here

Intel to ship new Atom server processors next week

Sorry, I could not read the content fromt this page.

View the original article here

Gentlemen, start your solid-state engines: Intel takes overclocking to SSDs

If cash is tight, upgrading to a solid-state drive is easily one of the biggest PC bangs you can get for your bottom dollar. Seriously, these puppies are face-melting fast, and the cheapest SSD you can buy blows the spinning platters off the fastest traditional hard drives you can buy.

But what if that's still not fast enough? What if you've got a need—a need for truly blistering storage speed?

Intel just might have you covered, oh impatient one. At the company's annual Intel Developer Forum this September, Intel plans to demonstrate a way to overclock its self-branded SSDs, as part of a larger overclocking session that mostly looks devoted to Haswell and upcoming Ivy Bridge-E Extreme Edition processors.

While overclocking processors, graphics cards, and even RAM is all relatively commonplace, overclocking SSDs is virtually unheard of—but that doesn't mean it isn't theoretically possible. Solid-state drives rely on NAND flash memory, and like those other frequently tweaked PC components, NAND is regulated by clock speeds and other technical fineries, all of which are regulated by the firmware in your drive's SSD controller.

When the upcoming SATA Express spec rolls out with 2GBps speeds, though, overclocking SSDs should be a much more practical performance option for PC enthusiasts.

Myce.com claims to have found hints about SSD overclocking abilities while digging around in preliminary code for Intel's XTreme Tuning Utility, a tool used to overclock and monitor your system hardware. Myce's sleuthing revealed options to control both the clock speed and the NAND bus speed of SSDs. The site also discovered "limited," "typical," and "unrestrained" power-mode options.

The scarcity of solid-state overclocking tools isn't necessarily a bad thing, though. Any time you overclock a component, you run the risk of stability issues—and an unstable storage drive is no storage drive at all. You also have to wonder how overclocking will affect the longevity of solid-state drives, especially as the industry adopts SSDs using NAND flash with triple-level cell technology, which (theoretically) sport shorter life spans than SSDs that use one or two cells.

In other words, drive reliability could be a big concern when overclocking SSDs.

Assuming the overclocking abilities come via the Xtreme Tuning Utility, Intel presumably won't let you push hard enough to fry your drive. Even still, there's probably no real reason to overclock SSDs—yet.

X-bit Labs, which first noticed the IDF session details, points out that the storage speed bottleneck in enthusiast PCs isn't so much the speed of the SSD itself as much as the 0.6 gigabyte per second (GBps) cap on the SATA 3.0 interface. Overclocking an SSD that already saturates the SATA interface is pretty much pointless.

When the upcoming SATA Express spec rolls out with 2GBps speeds, though, overclocking SSDs should be a much more practical performance option for PC enthusiasts with a need for ludicrous speed. Stay tuned for all the details out of IDF, which will be held September 10 - 12 at the Moscone Convention Center in San Francisco.

Brad Chacos spends the days jamming to Spotify, digging through desktop PCs and covering everything from BYOD tablets to DIY tesla coils.
More by Brad Chacos

View the original article here

Intel expands custom processor business

Intel is putting a sharp focus on expanding its custom processor and chip operations in response to a growing trend of companies building servers in-house to meet specific workloads or data center designs.

Last year, Intel has supplied custom processors to 18 companies, most notably eBay and Facebook, said Diane Bryant, senior vice president and general manager of Intel's Datacenter and Connected Systems Group, in an interview.

"That trend is growing. In the last year we've delivered 18 custom silicon processor solutions for the full array of customers -- our direct customers, the OEMS and the end users -- in order to meet their specific needs," said Bryant, who runs Intel's most profitable group.

The growing custom processor business will supplement Intel's bread and butter server chip business, consisting of selling generic Xeon processors available in rack, tower and blade servers sold by companies like Dell and Hewlett-Packard. But server infrastructures are changing with the growing adoption of cloud computing, big data and other applications, which has translated into a growing demand for custom processors, Bryant said.

Companies like Facebook and Google with mega data centers design servers in-house, and get them made from direct server suppliers like Quanta. The barebones servers typically cut excess components and are good enough to handle the growing cloud transactions such as search requests and social networking tasks. Facebook and Google have experimented with ARM processors in servers, while Tilera processors have also been tested.

"When you work with these end users who have technology as their core business, they are very clear on what is required," Bryant said. "They know what their workloads are, what their various applications are, they know what metric they are looking to hit from a performance per total cost of ownership."

The level of processor and chip customization varies with the workload, data center design, and even cooling solutions. Bryant provided an example where a flexible cooling system in a data center would allow customers to run processors at a higher frequency.

"We will have customers that have a very [specific] power target, so we will create versions whether its through changes in frequency, changes in core count, changes to drive down the power," Bryant said.

Customers usually give information about the applications they are running, the accelerators they need, the performance and power consumption levels they are looking to hit. Intel then customizes processors and chips that meets the specifications. Some customers in the technology and the data center business get specific about the server infrastructure.

"It all boils down to scale. I had one cloud service provider who had told me a single application is running across tens of thousands of servers. You can afford to tune that server very targeted against that application and eke out every bit of performance at ever lower cost of operations," Bryant said.

Intel is also now able to build system-on chip (SoC) designs, in which the CPU is combined with other accelerators, I/O, graphics and other processing units. That makes it easier to build custom processors and chips, Bryant said.

"With our SoC capability now, we can actually do rapid turns of our base product with very unique accelerators. Whether it's voice recognition acceleration or encryption or graphics acceleration... all the different types of accelerators that are targeted at different apps. We can deliver unique products there too," Bryant said.

Intel next year will release Xeon server chips based on the Broadwell processor core, which will succeed Haswell. Bryant said that the server SOC will also help optimize the chip to workloads, be it analytics or cloud.

"We have this wonderful Xeon core, and now Intel has a system-on-a-chip capability where we can rapidly turn out grabbing different intellectual property blocks and accelerators. Why not take this Xeon core and marry it with the SoC capability, and come up with... very [specific] processors targeted at unique capabilities," Bryant said.

In some ways, Intel is taking the same route as Advanced Micro Devices, which is creating custom chips based on its CPU and graphics architectures, but largely for non-server products. AMD's custom chips will be used in the upcoming Sony PlayStation 4 and Microsoft Xbox One gaming consoles.

Intel is also investing in software development to tie applications directly to chip development. The chip maker has released its own version of Hadoop, and is also actively contributing an orchestration layer to OpenStack so resources are effectively allocated at server, storage and network levels in distributed computing environments.

Beyond the server, Intel is also looking to change data center design. One of the projects called Rack Scale aims at decoupling the processing, I/O and storage units in data centers with faster throughput mechanisms.

"Instead of a rack being 24 servers slotted in, with each of those servers with compute, memory and I/O, instead break that artificial barrier of the server down and look at it at the rack level. And create pools of compute, memory and I/O so that the application can access and use whatever capacity it needs," Bryant said.

The company is expected to announced a new optical throughput standard called MXC, which will be detailed at the Intel Developer Forum next month. The company is also developing processors for different target markets, Bryant said. Intel will also announce a new Atom processor called Rangeley for embedded networking devices in early September, ahead of IDF.

"We have hundreds of microprocessor products to cover the entire space," Bryant said.

Agam Shah covers PCs, tablets, servers, chips and semiconductors for IDG News Service. Follow Agam on Twitter at @agamsh. Agam's e-mail address is agam_shah@idg.com

View the original article here

Intel bringing vision, 3D to laptop and tablet cameras

From mundane 2D devices, integrated cameras in laptops and tablets in the future will change into powerful 3D tools that can sense movement, track emotion, and even monitor reading habits of children, according to Intel.

Intel is developing a "depth sensing" camera, which is an enhanced version of a 3D camera that can go deeper inside images to "bridge the gap between the real and virtual world," said Anil Nanduri , director of perceptual products & solutions at Intel.

The webcam enhancements will help the computer understand a human better, bring new levels of interactivity to 3D games, and make webconferencing fun by blanking out the background and adding a green screen, Nanduri said.

"You'll add the ability to sense your excitement, emotion -- whether you are happy or smiling. The algorithms and technologies are there, but they are getting more refined, and as they get more robust, you'll see them," Nanduri said.

Such depth cameras will be integrated into laptops and ultrabooks in the second half of 2014. The technology will initially appear in external webcams such as Senz3D external webcam, which was jointly developed by Logitech and Intel, and will become available in the coming quarters. The camera technology will ultimately trickle down to tablets and smartphones, Nanduri said.

The camera will also be able to identify characteristics, contours and shapes of items in view. For example, the camera's ability to sense distance, size, depth, color, contours and other parameters of structures could also help in the growing area of 3D printing. A depth sensing picture of a model can be extracted to reveal exact specifications and other details of a design, which can then be printed.

"You are not going to look for a case [for a device] anymore, you'll just point that device, and the cameras will recognize what you have. It'll know the model number...and it'll print [the case] for you, or you go to the store, they will print it for you," Nanduri said.

With the help of eye tracking, it could also track how well somebody is reading and use that information as an evaluation tool. For example, it could track reading, and tell if kids are stuck on words, how much they read, or whether they need help with specific words.

"Having the capabilities to say -- they read about 80 percent of the lines, they had difficulty with these words -- that kind of intelligence for educational tools is phenomenal," Nanduri said.

Other small enhancements also include using a motion-sensing game where hands can be followed to pick up objects in the wider dimension of 3D games. The data collected by the camera could be combined with other modalities like voice recognition to improve human-computer interaction.

There are already 3D cameras out there, but Intel is trying to tack on the algorithm and hardware features that make images more meaningful.

"Kinect was a good initial version of a depth camera more from a long range perspective. When Intel started looking at it, we were primarily looking at it primarily as more personal interaction, short range, which is probably a meter or meter-and-a-half range of interaction," Nanduri said.

Integration in the thin ultrabook display panels may be a challenge. Intel is addressing the challenges with a high resolution short-range camera that focuses on a small area, and what Nanduri called "finger-level articulation."

"You need to have a lot more resolution for that zone. To really scale it to volumes, you need to get to the right form factor from the optics perspective, you need to get to the right power levels and you need to have the right cost structure to help scale it into integration," Nanduri said.

When the technology reaches devices, users may progressively forget the keyboard and mouse when interacting with computers.

"When you have depth information, what you can do with it is pretty phenomenal," Nanduri said.

Agam Shah covers PCs, tablets, servers, chips and semiconductors for IDG News Service. Follow Agam on Twitter at @agamsh. Agam's e-mail address is agam_shah@idg.com

Agam Shah is a reporter for the IDG News Service in New York. He covers hardware including PCs, servers, tablets, chips, semiconductors, consumer electronics and peripherals.
More by Agam Shah, IDG News Service

View the original article here

Intel set to ship LTE chip for mobile use in many countries

Intel plans to become a leader in wireless and will ship its first multimode LTE processor later this month as part of that drive.

The XMM 7160, a modem chip for mobile devices, can be equipped to work on as many as 15 LTE spectrum bands as well as on 2G and 3G networks. Multimode capability between 2G, 3G and 4G is a necessity for smartphones because most operators with LTE still rely on older technologies in parts of their service areas, and for voice calls. Intel discussed the chip earlier this year but announced on Monday that it will ship this month.

Intel is still an underdog in the mobile industry but sees itself as one of just a handful of potential rivals to the dominant Qualcomm. At the core of Intel's mobile device push is the former Infineon wireless business, which it acquired in 2010.

While Infineon was a "fast follower" in mobile, intentionally timing its products for when commercial volumes were starting to ship, it's been moving toward market leadership since the acquisition, said Hermann Eul, vice president and general manager of Intel's mobile and communications group.

"For us, it is the clear strategy, the clear plan, to go into a leadership position, and we are currently in the transition to do that," Eul said. He spoke on Monday at a press and analyst briefing at Intel headquarters.

Intel painted itself as a force that can make the mobile chip industry more competitive and help to drive technology forward. Device makers, mobile operators and other players are counting on that, according to Aicha Evans, vice president and general manager of wireless platform R&D.

"This is the first time that I've been at Intel working at anything, that the industry is actively rooting for us to succeed ... because of diversity in this market," said Evans, who has been at Intel for seven years.

Qualcomm already sells multimode LTE chips for a large number of bands, but Intel claims the 7160 has a performance edge over the competition. It's 12 percent smaller than rival products and consumes 20 percent to 30 percent less power, the company said.

Support for multiple LTE bands is important because there are about 40 different spectrum bands allocated for the high-speed mobile data system around the world. Phones may need many different bands in order to allow international roaming on LTE, and being able to pack in many frequencies can also save device makers money.

With the 7160, a device maker could have a single chip for phones sold all over the world, said Thomas Lindner, senior director in the mobile and communications group. The 15 bands that it can support are not set in stone but can be configured for particular manufacturers, he said.

The first iteration of the 7160 will be equipped for so-called Category 3 LTE, with throughput as high as 100Mbps (bits per second). An updated version shipping by year's end will offer Category 4 LTE, which can go as high as 150Mbps. Speeds for individual users won't match those maximums, but each advance should mean better real-world performance. The second version of the 7160 will also support VoLTE (voice over LTE), which will let carriers shift voice calls from their 2G and 3G systems onto what are now data-only LTE networks.

The chip could also be set up to support TD (time-division) LTE, a form of the technology that Sprint, China Mobile and other carriers plan to use. Whether Intel implements TD-LTE in the 7160 will depend on demand from operators, Evans said.

In parallel with the 7160, Intel has been developing the XMM 7260, which is due to ship in the first half of next year. The 7260 can support carrier aggregation, an emerging LTE feature that lets operators combine two separate spectrum bands into one for higher performance. On Monday, Intel demonstrated carrier aggregation on a 7260 evaluation board built at its new design center in San Diego, Qualcomm's hometown. The 7260 also will include TD-LTE.

Intel has a major potential advantage over wireless rivals in the form of its advanced chip design expertise and private foundry capacity, said analyst Jack Gold of J. Gold Associates, who attended the event. This is the chipmaking operation that churns out smaller, faster, more efficient microprocessors on a steady schedule. Qualcomm, by contrast, relies on outside manufacturers.

The former Infineon wireless business is still relying on TSMC (Taiwan Semiconductor Manufacturing Corp.), but Intel expects to bring its own chipmaking assets to bear on the wireless business within two to three years, Evans said. The sooner it can do so, the more benefit it can gain, Gold said.

Stephen Lawson covers mobile, storage and networking technologies for The IDG News Service. Follow Stephen on Twitter at @sdlawsonmedia. Stephen's e-mail address is stephen_lawson@idg.com

View the original article here

Intel previews multimode LTE chips, with eyes on U.S. handset market

Intel will ship multimode LTE capabilities by the end of the month, Intel executives said Monday. It’s a technology the company is throwing its weight behind—and needs, if it’s to compete in the U.S. market.

Why? Because LTE is a pitch right in Intel’s wheelhouse—high-speed mobile data makes the company’s CPUs and GPUs even more valuable, while the relatively high power consumption those chips require can be mitigated with Intel’s process technology. And LTE has been the gating factor in placing Intel’s mobile chips into U.S. handsets.

Mark HachmanHermann Eul, the general manager of the Intel Mobile and Communications Group, shows off the XMM 7160 at a press event Monday.

Intel began shipping a single-mode LTE solution in the fourth quarter of 2012, and it plans to ship a multimode chip, the XMM 7160, by the end of the month, Aicha Evans, the vice president of Intel’s Mobile and Communications Group, said in a press conference on Monday.

At Intel, combining communications with mobile computing is inseparable: “Everything that computes must connect,” Evans said.

As the name implies, single-mode LTE simply provides LTE connectivity—great where LTE connectivity is relatively available, but with no provision to jump to other bands if the user roams between networks. With multimode LTE technology, the chips can support a number of LTE networks around the world, providing range. (If a LTE network is not present, the XMM 7160 can “back down” to older, 3G technologies.)

Specifically, the XMM 7160 will support 15 LTE bands, with 20 to 30 percent lower power consumption than competitive solutions. The XMM 7160 will also be about 12 percent physically smaller than competing chips, Evans said. Both Qualcomm and Intel are shipping or nearly shipping multimode LTE; MediaTek, Broadcom, Marvell, and Nvidia have announced multimode LTE plans.

The problem that Intel faces—and that the chipmaker is hurriedly working to overcome—is the reputation as a “fast follower” in the wireless space. Intel bought Infineon’s wireless business in 2010, for $1.4 billion. But at the time of that purchase, Infineon’s technology was only at the 2G/3G stage. Since then, the company has worked to catch up, running design teams in parallel “to step up to a leadership position,” said Hermann Eul, the general manager of the Intel Mobile and Communications Group.

Intel isn’t there yet. In June, for example, rival Qualcomm announced the integration of multimode 3G/4G LTE into its Qualcomm 400 processors. That’s a step beyond Intel; Intel’s XMM 7160 is a discrete chip, and the company has not announced plans to integrate the XMM 7160 functionality into its application processors, Eul said. “We will do it when the time is right,” he added.

Intel’s announcement, as well as its upcoming integration plans, represent a key strategic element for Intel’s continued push into the wireless business. Intel has expended a great deal of effort to push its mobile processors—the “Clover Trail+” Atom chip, and the upcoming “Bay Trail”—into mobile handsets. But its only design win of note is the Lenovo K900 handset, restricted to the Chinese market.

Intel's new XMM 7160 LTE chip.

Why aren’t Intel’s mobile chips in handsets selling within the U.S.? “Absence of LTE is the reason,” Tom Kilroy, Intel’s vice president of sales, said, according to Engadget. “We can’t get ranged by U.S. carriers without LTE, so once we have multimode LTE coming to market later this year, we have an opportunity to compete in that business.”

According to Will Strauss, a digital signal processor analyst with Forward Concepts, Intel’s XMM 7160 chip has been qualified with AT&T and Verizon. Evans declined to comment, but acknowledged that the lack of LTE has held the company back.

Finally, there's the computing element. Intel executives said that download speeds would be up to 100 Mbit/s, a theoretical limit that's dependent on the carrier and available spectrum. But without that additional data, it's difficult to justify a premium Intel CPU versus a cheaper model.

Now, Intel simply has to convince handset makers that it's a player. In 2014 and 2015, Intel is striving for “credibility,” Evans said, while its technical roadmap includes such capabilities as voice over LTE and LTE Advanced technology. A 7260 chip may include those capabilities.

View the original article here

Intel to customize chips for big data applications

Intel, increasingly customizing server chips for customers, is now tuning chips for workloads in big data.

Software is becoming an important building block in chip design, and customization will help applications gather, manage and analyze data a lot quicker, said Ron Kasabian, general manager of big data solutions at Intel.

Through hardware and software improvements, the company is trying to figure out how its chips can perform better in areas like predictive analytics, cloud data collection and specific task processing. The company has already released its own distribution of Hadoop, a scalable computing environment that deals with large data sets, and now chip improvements are on tap.

Kasabian said Intel is starting with the software. “It takes a while to get silicon to market,” he said. “We understand where we can optimize for silicon, and there are certain things to [improve] for performance and optimization.”

The company is taking lessons from software implementations and then looking to enhance the silicon to fill any software gap, Kasabian said, adding that the chip-design process takes about two years.

Server makers have been customizing servers specifically to carry out big data workloads, and improvements at the chip and instruction-set level could speed up task execution.

The plan includes developing accelerators or cores for big-data type workloads. For example, Intel is working with Chinese company Bocom to implement the Smart City project, which tries to solve counterfeit license plate problems in China by recognizing plates, car makes and models. The project involves sending images through server gateways, and Intel is looking to fill software gaps by enhancing the silicon. One improvement could be implementing accelerators to decode video, Kasabian said.

Intel has a big software organization, and the appointment of Renee James—formerly head of the software unit—earlier this year as the company’s president was a sign of the chip maker’s intent to dig deeper into software. The company does not want to become a packaged software distributor, but wants to enable software to work better on Intel architecture hardware. For a long time, Intel has backed open-source software and has hundreds of coders contributing to the development of Linux.

Different industries have different implementations of big data, Kasabian said. For example, a big data problem in genomics could differ from one in telecommunications.

Intel is also entering the space of the Internet of things, an emerging field in which networked devices with embedded processors and sensors are used as data-gathering instruments. The company has assets such as McAfee’s software and hardware platform and Wind River’s real-time operating system for its embedded chips to quickly process and securely collect data.

Outside of the silicon, Intel is focusing on providing the right software tools for data centers. Hadoop was the starting point, and now Intel is looking closely at analytics, Kasabian said.

Attaching Intel’s name to Hadoop will “kind of ease the mind of folks in enterprises,” Kasabian said, adding that implementation of the platform in data centers will be easier.

A lot of research is also taking place at Intel labs on stream processing and graph analytics as the company designs chips and tweaks software.

“We’re looking at all the big industry categories,” Kasabian said.

Agam Shah is a reporter for the IDG News Service in New York. He covers hardware including PCs, servers, tablets, chips, semiconductors, consumer electronics and peripherals.
More by Agam Shah, IDG News Service

View the original article here

Wi-Fi adapter shootout: Qualcomm versus Intel in an 802.11 battle

Gamers are always hunting for a competitive edge, and the folks at Bigfoot Networks—now a part of Qualcomm Atheros—have long promised to deliver network interface cards that perform better with online games and other latency-sensitive applications. To demonstrate its prowess in this area, the company sent me two identical Alienware notebooks, one equipped with Qualcomm’s Killer Wireless-N 1202 and the other with Intel’s Centrino Advanced-N 6230.

Robert CardinWe benchmarked two otherwise identical Alienware laptops: one outfitted with a Killer Wireless-N 1202 NIC (left) and the other with Intel's Centrino Advanced-N 6230 (right).

Both NICs are dual-band adapters that can connect to an 802.11n router on either the 2.4GHz or 5GHz frequency band. Both also support two spatial streams for a maximum physical link rate of 300 megabits per second. Some gaming-laptop manufacturers, including Alienware, offer Killer NICs as standard equipment, while others offer the adapters as added-cost upgrades. You can also purchase one of these cards by itself and upgrade your existing notebook, provided that the system has an available Mini PCIe slot to host the card (a common feature on better notebooks). The Killer Wireless-N 1202 is certainly inexpensive enough: I’ve seen it selling online for as little as $35 (Intel’s card is street-priced at about $30).

The key selling point of Killer NIC technology is its ability to identify the types of traffic traveling over your network and to assign higher priority to latency-sensitive traffic, such as online games, HD video, and audio.

Latency is a measure of time delay. When applications such as online games and streaming media encounter too much latency, you’ll end up with visible and/or audible glitches and hiccups. If you’re playing a first-person shooter with an online opponent, latency can render you a frustratingly easy target.

To evaluate each card’s ability to combat network latency, Qualcomm provided me with its Gaming Network Efficiency (GaNE) benchmark to measure ping (the time required for a packet to make a round trip on the network) and jitter (undesirable deviations in signal timing). Qualcomm offered to allow us to examine the program’s source code to ensure that there were no shenanigans.

GaNE measures real-time performance from two wireless clients at once, recording the results on a third computer that’s hardwired to the network. This way, both client adapters are subject to the same environmental conditions—an important variable when you’re benchmarking wireless performance. The tool measures latency between two networked PCs by sending a 100-byte packet on a round trip over the network every 50 milliseconds (100 bytes is the typical packet size on gaming networks, and 50ms is the typical interval between packets on the same). I used my longtime favorite 802.11n Wi-Fi router, a dual-band Asus RT-N66U, for these tests.

The GaNE benchmark, which measures latency and jitter, shows the Killer NIC to be the superior Wi-Fi adapter for latency-sensitive applications such as online games.

I set up the two wireless clients about 9 feet from the router in the same room, so no walls stood in between, and I connected them to the router’s 2.4GHz network. I then performed six runs of the GaNE benchmark. GaNE reported an average ping time for the Killer Wireless-N 1202 of about 1.5 milliseconds, and an average ping of nearly 4 milliseconds for the Intel card. What’s more, GaNE measured one-and-a-half times more jitter with Intel’s card than it did with Qualcomm’s. The Killer NIC delivered similar performance when I switched the clients over to the router’s 5GHz network—slightly less than 2 milliseconds—but Intel’s card registered a much higher average ping of nearly 8 milliseconds.

I also used the Asus RT-N66U to compare the adapters’ TCP-throughput performance, using JPERF (the Java front end to the TCP-throughput benchmark iPERF). The two adapters traded places on the 2.4GHz frequency band, but when the Killer Wireless-N 1202 won, it won by a significant margin. When the client was in the same room as the router, 9 feet away with no walls in between, it was 10.7 mbps faster than Intel’s component. And when the client was in my home office, 65 feet away and separated by three insulated interior walls, the Killer NIC was more than 25 mbps faster. At the two locations at which Intel’s adapter prevailed—in the kitchen (20 feet from the router) and in the home theater (35 feet from the router)—the margins were just 0.5 mbps and 4.0 mbps respectively.

Qualcomm's Killer Wireless-N 1202 and Intel's Advanced-N 6230 had comparable performance on the 2.4GHz band, but the Killer NIC was much faster when situated far from the router.

When I tested each adapter on the 5GHz network, Intel’s NIC performed much better—at close range, at least. In the bedroom test, the Centrino Advanced-N 6230 delivered TCP throughput of 196.0 mbps to the Killer’s 121.0 mbps. And when I moved the clients to the kitchen, Intel’s component was more than twice as fast as Qualcomm’s. But when I tested each client at longer distances, the tables turned, and the Killer NIC delivered dramatically higher TCP throughput than Intel’s card. In the home theater, the Killer delivered 81.1 mbps to the Centrino’s 30.1 mbps; and in the home office, Qualcomm’s card produced a rate of 63.8 mbps to Intel’s 34.9 mbps.

On the 5GHz band, Intel's adapter was much faster than the Killer NIC at close range. But the Killer card turned the tables at long range.

So Qualcomm’s Killer Wireless-N 1202 is a great Wi-Fi adapter. What happens if you’ve upgraded—or are planning to upgrade—to a router based on the new 802.11ac standard? Well, such routers are backward-compatible with 802.11n, so a Killer NIC will still work with one. But if that’s where you’re at, I think the better bet is to switch to an 802.11ac adapter. In fact, some notebook manufacturers—including Alienware—are already offering 802.11ac Wi-Fi adapters as standard equipment on higher-end laptop models (and tellingly, Alienware does not offer the Killer NIC—which is currently available only in 802.11n configurations—as an option on those models).

You should also remember your other option: a hardwired ethernet connection. If you don’t mind stringing cable, a hardwired connection will be faster than any wireless setup, and you won’t have to worry much about latency or jitter at all. But most people don’t want to hassle with cables these days, because wireless connections are so much more convenient.

If you’re buying a new gaming notebook and the seller offers the Killer NIC as a modestly priced upgrade (say, no more than $30), it’s definitely worth the money. The same goes if you’re upgrading a laptop with a Mini PCIe expansion slot and two internal antennas (Qualcomm's Killer Wireless-N1103 supports three antennas). But if you can get an 802.11ac adapter, that would be an even better investment. In fact, I’m surprised Qualcomm isn’t already offering a Killer NIC based on that spec.

A lapsed musician, Michael is passionate about home and personal audio. He uses his home as a real-world test lab for evaluating consumer electronics and home control technology.
More by Michael Brown

View the original article here