SAN FRANCISCO – Eighteen months after Sun Microsystems and Intel made peace and announced plans for Intel-based servers as well as working together on Intel-optimized software, the two companies held a briefing with reporters here Tuesday to update their progress and future direction.
Much of their work has centered around optimizing OpenSolaris, the experimental version of Solaris (Sun's Unix variant) where new features are tested and debugged before being added to the official Solaris product that is shipped with Sun servers. That deal is bearing fruit already in helping Sun (NASDAQ: JAVA) optimize Java for Solaris.
"The January '07 deal gained us access to Intel's architecture to do things in Solaris we could not do before," said Herb Hinstorff, director of marketing for Sun Solaris. "The first year was all about getting to know each other. This year, we are shipping Intel optimizations in Solaris now and Intel is a contributing member to OpenSolaris."
Dave Stewart, a software engineering manager at Intel (NASDAQ: INTC), added: "It has been tremendous to see the results of the collaborative effort of marrying Solaris with Xeons. We're now working with the Sun xVM team to deliver some virtualization optimizations."
Beyond the CPU
But the work isn't just on the CPU level. Andy Roach, senior director of x64 engineering for Solaris, added "We're working on more than just CPUs. We're working to deliver full Intel-based solution stacks to Solaris users." He said Intel developers are frequently seen in Sun offices and vice versa.
Sun gives away OpenSolaris and new versions are released every six months, as opposed to the multi-year gap between Solaris releases. After the developer community has had a chance to experiment with technologies on OpenSolaris, such as virtualization, power management and CPU optimization, those technologies will migrate to the enterprise product.
The two companies made Penryn optimizations to OpenSolaris and Hinstorff said they hope to have all of the enabling technologies for Nehalem, Intel's new processor architecture due by the end of the year, in the next release of OpenSolaris.
There will have to be optimizations because Nehalem will represent a major change in architecture over the current Xeon design. The memory controller will be on the CPU, noted Roach, which will mean a major drop in memory latency and improvement in performance. Nehalem will also use a different kind of memory, DDR3.
The two companies are also working on Java performance tuning. Within the first six months of the alliance by the two firms, Java performance had improved 20 percent, and at JavaOne earlier this year, they announced a 68 percent benchmark improvement. The Java groups at Sun and Intel are working on optimizations across multiple operating systems, with the goal "to have best performing Java on Intel as possible," said Hinstorff.
IDC vice president Jean Bozman said the alliance is definitely starting to benefit Sun. "Sun has seen an improvement in its x86 sales, but there are multiple reasons for it," she told InternetNews.com. "One reason is they are now able to tap into a very large segment of the marketplace where they hadn't been participating before. By optimizing Solaris for Intel, it could only make things better for them."
Saturday, August 9, 2008
HP, Intel, Yahoo Team on Cloud Computing Labs
Is there gold in them thar clouds? Yahoo, Intel and HP seem to think so.
The trio of tech titans announced an ambitious research initiative today focused on studying the software, hardware and datacenter management issues surrounding cloud computing. The group's new Cloud Computing Test Bed is aimed at creating a large, globally distributed testing environment that they hope will encourage unprecedented levels of research.
Companies from Amazon to Google, IBM and Salesforce.com already offer some cloud computing (define) services, but Prith Banerjee, senior vice president of research at HP and director of HP Labs, said a larger effort is needed.
"It requires an entirely new approach to the way we manage, and deploy cloud computing," he said in a conference call with reporters.
Banerjee also said the participation of HP Labs fit with the group's renewed commitment to focus more on projects that have a clear commercial payoff.
Other partners for the test bed include Infocomm Development Authority of Singapore (IDA), the University of Illinois at Urbana-Champaign, and Germany's Karlsruhe Institute of Technology (KIT). The partnership with the University of Illinois also includes a grant from the National Science Foundation.
The test bed will initially consist of six locations at IDA, the University of Illinois, the Steinbuch Centre for Computing at KIT, HP Labs, Intel Research and Yahoo.
The companies said each location will host a cloud computing infrastructure, with HP and Intel providing the hardware and processors. Each center will have 1,000 to 4,000 processor cores.
All six locations are expected to become fully operational later this year, when they'll also become available to researchers through a worldwide selection process.
"We really need to think of this beyond the physical hardware layer," said Prabhakar Raghavan, head of Yahoo Research. "It's about what exciting applications you can build once you take the cloud for granted. What if a utility was always available at any location or scale we want?"
Gartner analyst David Mitchell Smith said the research effort could potentially pay off by providing companies and organizations with ways to use cloud services to offload the huge cost of buying and maintaining datacenters.
"People are starting to imagine the cloud on that scale -- it's not pie in the sky anymore," Smith told Internetnews.com.
Intel (NASDAQ: INTC) Research Director Andrew Chien said the scale of project will help to expose what works and what needs more research. "It's one thing to do things at a test tube level, it's another to operate at a much larger scale with network effects and contention," he said.
Cloud computing research on a large scale is far from new. Last year, Google (NASDAQ: GOOG) and IBM (NYSE: IBM) also announced a joint project in conjunction with several universities. When asked why Intel and its partners didn't simply join that effort, Chien said he thought his group's effort is both complementary and different than the Google/IBM project.
"What we're trying to do is support research in a variety of levels and novel hardware features Intel has been able to add in silicon," he said. "We'll allow people to run a fairly low level of customized software."
Gartner's Smith said the reason Yahoo, Intel and HP are going their own way is simple: "They're competitors."
Additionally, he doesn't think a broader collaboration of competitors is likely in the near future.
"I wouldn't expect one unifying 'cloud' vision anytime soon," he said. "You'll see different definitions by different vendors."
The trio of tech titans announced an ambitious research initiative today focused on studying the software, hardware and datacenter management issues surrounding cloud computing. The group's new Cloud Computing Test Bed is aimed at creating a large, globally distributed testing environment that they hope will encourage unprecedented levels of research.
Companies from Amazon to Google, IBM and Salesforce.com already offer some cloud computing (define) services, but Prith Banerjee, senior vice president of research at HP and director of HP Labs, said a larger effort is needed.
"It requires an entirely new approach to the way we manage, and deploy cloud computing," he said in a conference call with reporters.
Banerjee also said the participation of HP Labs fit with the group's renewed commitment to focus more on projects that have a clear commercial payoff.
Other partners for the test bed include Infocomm Development Authority of Singapore (IDA), the University of Illinois at Urbana-Champaign, and Germany's Karlsruhe Institute of Technology (KIT). The partnership with the University of Illinois also includes a grant from the National Science Foundation.
The test bed will initially consist of six locations at IDA, the University of Illinois, the Steinbuch Centre for Computing at KIT, HP Labs, Intel Research and Yahoo.
The companies said each location will host a cloud computing infrastructure, with HP and Intel providing the hardware and processors. Each center will have 1,000 to 4,000 processor cores.
All six locations are expected to become fully operational later this year, when they'll also become available to researchers through a worldwide selection process.
"We really need to think of this beyond the physical hardware layer," said Prabhakar Raghavan, head of Yahoo Research. "It's about what exciting applications you can build once you take the cloud for granted. What if a utility was always available at any location or scale we want?"
Gartner analyst David Mitchell Smith said the research effort could potentially pay off by providing companies and organizations with ways to use cloud services to offload the huge cost of buying and maintaining datacenters.
"People are starting to imagine the cloud on that scale -- it's not pie in the sky anymore," Smith told Internetnews.com.
Intel (NASDAQ: INTC) Research Director Andrew Chien said the scale of project will help to expose what works and what needs more research. "It's one thing to do things at a test tube level, it's another to operate at a much larger scale with network effects and contention," he said.
Cloud computing research on a large scale is far from new. Last year, Google (NASDAQ: GOOG) and IBM (NYSE: IBM) also announced a joint project in conjunction with several universities. When asked why Intel and its partners didn't simply join that effort, Chien said he thought his group's effort is both complementary and different than the Google/IBM project.
"What we're trying to do is support research in a variety of levels and novel hardware features Intel has been able to add in silicon," he said. "We'll allow people to run a fairly low level of customized software."
Gartner's Smith said the reason Yahoo, Intel and HP are going their own way is simple: "They're competitors."
Additionally, he doesn't think a broader collaboration of competitors is likely in the near future.
"I wouldn't expect one unifying 'cloud' vision anytime soon," he said. "You'll see different definitions by different vendors."
Intel Defies Economy, Seasonal Weakness
Intel on Tuesday announced a 25 percent jump in profits over the second quarter of 2007 thanks to its aggressive cost cutting and continued strength of its products, despite U.S. economic problems and seasonal softness.
Normally, the second quarter is the weakest time of year in the hardware business, but Intel sales rose 9 percent over 2Q07 to $9.5 billion for its fourth record quarter of revenue in a row. The chip giant's aggressive program of cost cutting has paid off to the tune of a 67 percent improvement in operating income and a net profit of $1.6 billion, a 25 percent improvement over last year.
That translates to 28 cents per share, three cents better than analysts were expecting. Intel shares rose 21 cents in after-hours trading, after rising 24 cents to $20.71 during the regular trading session of the day.
During a conference call with financial analysts, CEO Paul Otellini was very upbeat on the quarter's numbers. "Demand continues to be strong, with revenue and unit shipments at high end of the norm when taking into account the divestiture of the NOR business," he said.
Intel is in the process of finalizing the spin-off of its NOR RAM business with STMicroelectronics.
The 45nm manufacturing process continues to ramp up, with the company expected to ship 100 million chips this year, Otellini said. They are getting better yields than at 65nm and expect 45nm to become the dominant manufacturing process this quarter.
Another segment growing faster than Intel expected: notebooks. Otellini said that notebook processor sales surpassed desktops in Q2. "That was sooner than we expected," he said.
The other surprise is Atom, its embedded chip. Atom sales are exceeding expectations and unit shipments will grow sharply in second half. The company is increasing production of Atom every 40 days as demand rolls up in netbook, embedded devices and consumer electronics.
Chief Financial Officer Stacy Smith said that microprocessor revenue was up 14 percent, with mobility accounting for one-third of total quarterly revenue and up 15 percent from 2Q07. All geographies experienced year-to-year growth, including the U.S., and were a little better than average on a seasonal basis.
"We are aware of global issues that dominate the market these days," he said. "We saw order patterns [grow] as anticipated. Inventories are healthy, our global footprint is benefiting from demand."
For the third quarter, Intel projects revenue between $10 and $10.6 billion, which would be two percent growth year over year at the midpoint of that range. Gross margin will be around 58 percent
Normally, the second quarter is the weakest time of year in the hardware business, but Intel sales rose 9 percent over 2Q07 to $9.5 billion for its fourth record quarter of revenue in a row. The chip giant's aggressive program of cost cutting has paid off to the tune of a 67 percent improvement in operating income and a net profit of $1.6 billion, a 25 percent improvement over last year.
That translates to 28 cents per share, three cents better than analysts were expecting. Intel shares rose 21 cents in after-hours trading, after rising 24 cents to $20.71 during the regular trading session of the day.
During a conference call with financial analysts, CEO Paul Otellini was very upbeat on the quarter's numbers. "Demand continues to be strong, with revenue and unit shipments at high end of the norm when taking into account the divestiture of the NOR business," he said.
Intel is in the process of finalizing the spin-off of its NOR RAM business with STMicroelectronics.
The 45nm manufacturing process continues to ramp up, with the company expected to ship 100 million chips this year, Otellini said. They are getting better yields than at 65nm and expect 45nm to become the dominant manufacturing process this quarter.
Another segment growing faster than Intel expected: notebooks. Otellini said that notebook processor sales surpassed desktops in Q2. "That was sooner than we expected," he said.
The other surprise is Atom, its embedded chip. Atom sales are exceeding expectations and unit shipments will grow sharply in second half. The company is increasing production of Atom every 40 days as demand rolls up in netbook, embedded devices and consumer electronics.
Chief Financial Officer Stacy Smith said that microprocessor revenue was up 14 percent, with mobility accounting for one-third of total quarterly revenue and up 15 percent from 2Q07. All geographies experienced year-to-year growth, including the U.S., and were a little better than average on a seasonal basis.
"We are aware of global issues that dominate the market these days," he said. "We saw order patterns [grow] as anticipated. Inventories are healthy, our global footprint is benefiting from demand."
For the third quarter, Intel projects revenue between $10 and $10.6 billion, which would be two percent growth year over year at the midpoint of that range. Gross margin will be around 58 percent
Intel Shows Off New Centrino Notebook Chips
TAIPEI -- Intel, the world's top PC chipmaker, on Tuesday launched the next-generation of its Centrino wireless chip that it hopes will provide a new revenue stream amid a broader push into mobile technologies.
The launch of the Centrino 2 chip, previously codenamed "Montevina," came after a delay of several months and was decidedly lower key than the launch of the first Centrino chips in 2003.
The next-generation chipset combines Wi-Fi capability, which has a maximum range of only about 100 meters, with WiMAX, the newer wireless technology that allows for high-speed data transmission over much larger distances, and which can be used to blanket entire cities.
IDC analyst Bryan Ma said the new chips represent incremental development for the industry, compared with the first Centrino that marked Intel's (NASDAQ: INTC) entry to the wireless space.
"My big question is whether this is revolutionary or evolutionary; I suspect it may be more of the latter," Ma said. "Even if it's just evolutionary, however, it is still a good fuel to help the industry along."
The chips are mainly intended to go into notebook computers, as the PC industry moves to more mobile devices with new lighter technologies and development of new wireless networks.
Taiwan is in the process of constructing six WiMAX wireless networks, and is also the world's top contract PC producer, with sector leaders Quanta and Compal Electronics manufacturing for Hewlett-Packard (NYSE: HPQ), Dell (NASDAQ: DELL) and Sony (NYSE: SNE), among others.
"Notebook computers will be the main industry driver in the future, and notebook sales already outnumber desktop sales in many countries," said Stanley Huang, director of advanced technical sales and services for Intel Asia Pacific, at a launch event in Taipei.
"Because this chip has new capabilities, we hope it will change the way people think of mobile computing," he said.
Others at the launch included HP, the world's top PC maker, and Acer and Lenovo, the third- and fourth-place PC vendors.
All plan to use the chips in their models, with various players designing some 250 different notebook models with the chip.
The world's largest chipmaker is hoping to capitalize on the growing popularity of notebook PCs, which are rapidly taking over from older desktop models.
Data tracking firm IDC expects the notebook PC segment to grow 35 percent this year to 145 million units shipped, while desktops should grow much more slowly, by 2 percent, to 157 million. At current growth rates, IDC estimates annual notebook shipments will surpass desktops next year.
The Centrino 2 launch is part of a broader Intel strategy to develop a wider suite of wireless products to use in non-PC devices, most notably mobile phones, as data transmission speeds improve with new mobile technologies.
Such technologies allow for a much wider range of applications, such as streaming video and video downloads, that would have been impossible using older technology.
Paul Otellini, Intel's CEO, said in a speech last year that his company is seeking to spread its technology from the high-performance computing market to smaller products such as TV set-top boxes and handheld Internet-enabled devices.
The launch of the Centrino 2 chip, previously codenamed "Montevina," came after a delay of several months and was decidedly lower key than the launch of the first Centrino chips in 2003.
The next-generation chipset combines Wi-Fi capability, which has a maximum range of only about 100 meters, with WiMAX, the newer wireless technology that allows for high-speed data transmission over much larger distances, and which can be used to blanket entire cities.
IDC analyst Bryan Ma said the new chips represent incremental development for the industry, compared with the first Centrino that marked Intel's (NASDAQ: INTC) entry to the wireless space.
"My big question is whether this is revolutionary or evolutionary; I suspect it may be more of the latter," Ma said. "Even if it's just evolutionary, however, it is still a good fuel to help the industry along."
The chips are mainly intended to go into notebook computers, as the PC industry moves to more mobile devices with new lighter technologies and development of new wireless networks.
Taiwan is in the process of constructing six WiMAX wireless networks, and is also the world's top contract PC producer, with sector leaders Quanta and Compal Electronics manufacturing for Hewlett-Packard (NYSE: HPQ), Dell (NASDAQ: DELL) and Sony (NYSE: SNE), among others.
"Notebook computers will be the main industry driver in the future, and notebook sales already outnumber desktop sales in many countries," said Stanley Huang, director of advanced technical sales and services for Intel Asia Pacific, at a launch event in Taipei.
"Because this chip has new capabilities, we hope it will change the way people think of mobile computing," he said.
Others at the launch included HP, the world's top PC maker, and Acer and Lenovo, the third- and fourth-place PC vendors.
All plan to use the chips in their models, with various players designing some 250 different notebook models with the chip.
The world's largest chipmaker is hoping to capitalize on the growing popularity of notebook PCs, which are rapidly taking over from older desktop models.
Data tracking firm IDC expects the notebook PC segment to grow 35 percent this year to 145 million units shipped, while desktops should grow much more slowly, by 2 percent, to 157 million. At current growth rates, IDC estimates annual notebook shipments will surpass desktops next year.
The Centrino 2 launch is part of a broader Intel strategy to develop a wider suite of wireless products to use in non-PC devices, most notably mobile phones, as data transmission speeds improve with new mobile technologies.
Such technologies allow for a much wider range of applications, such as streaming video and video downloads, that would have been impossible using older technology.
Paul Otellini, Intel's CEO, said in a speech last year that his company is seeking to spread its technology from the high-performance computing market to smaller products such as TV set-top boxes and handheld Internet-enabled devices.
Intel Introduces First System on a Chip
As promised, Intel has jumped into the System-on-Chip (SoC) market with a series of announcements based around both the Atom processor and a new family, built on the Pentium M chip.
The company has 15 SoC projects in its pipeline, aimed at large and small form factors alike, from handheld devices and mobile Internet devices (MIDs) to cars to servers, all of them considered new, growth markets, according to Gadi Singer, general manager of Intel's SoC enabling group.
Intel (NASDAQ: INTC) sees its greatest opportunities in the mobile Internet space and in emerging markets. Singer predicted the Internet will have 1.2 billion users by 2012 and people will expect Internet connectivity wherever they go, said Singer. But mature markets want their Internet, too, having grown up with constant access.
"New devices need to be powerful, responsive, but in a small form factor," he told a briefing of journalists. "This puts a lot of pressure on the silicon to support this."
The company's first SoC product is the EP80579, a four-on-one chip based on the Pentium M processor design with memory controller, I/O controller and a set of integrated application-specific accelerators on a single chip. Doug Davis, vice president of the Digital Enterprise Group and general manager of the Embedded and Communications group, said Intel used the Pentium M because that's what it had available at the time development began and it met its needs.
According to Davis, the chip will be 45 percent smaller than if the four chips were installed separately on a motherboard and consume 34 percent less power as a whole rather than as four separate pieces.
The EP80579 will consume 11 to 21 watts, depending on clock speed, and run between 600MHz and 1.2GHz. To satisfy industrial requirements, the company has announced a seven-year support life cycle.
Another chip, the EP80579, is for the embedded market. The next generation will be based on Atom instead of Pentium M, hardly a problem since Atom is also based on the Intel Architecture (IA) instruction set. The next generation of the MID processor line will be Lincroft, due next year, featuring a tenfold reduction in power consumption over the current line of chips.
For consumer electronics (CE), the company plans to release a chip code-named Canmore later this year, followed by Sodaville next year. Intel did not go into detail beyond saying that the chips would be tuned for consumer electronics devices, such as bringing the Internet to the television.
IDC analyst Shane Rau said this is a big change for Intel. "These are very highly integrated processors, he told InternetNews.com. "Intel is known for stand-alone processors," he said. "What's different about these is there's a system on a chip. So Intel is prepared to introduce pieces of silicon that are app-specific."
This reflects Intel's desire to look for every way possible to expand, and the way it expands is by finding new markets to enter. "It's the PC market looking to redeploy its technologies into new markets,
The company has 15 SoC projects in its pipeline, aimed at large and small form factors alike, from handheld devices and mobile Internet devices (MIDs) to cars to servers, all of them considered new, growth markets, according to Gadi Singer, general manager of Intel's SoC enabling group.
Intel (NASDAQ: INTC) sees its greatest opportunities in the mobile Internet space and in emerging markets. Singer predicted the Internet will have 1.2 billion users by 2012 and people will expect Internet connectivity wherever they go, said Singer. But mature markets want their Internet, too, having grown up with constant access.
"New devices need to be powerful, responsive, but in a small form factor," he told a briefing of journalists. "This puts a lot of pressure on the silicon to support this."
The company's first SoC product is the EP80579, a four-on-one chip based on the Pentium M processor design with memory controller, I/O controller and a set of integrated application-specific accelerators on a single chip. Doug Davis, vice president of the Digital Enterprise Group and general manager of the Embedded and Communications group, said Intel used the Pentium M because that's what it had available at the time development began and it met its needs.
According to Davis, the chip will be 45 percent smaller than if the four chips were installed separately on a motherboard and consume 34 percent less power as a whole rather than as four separate pieces.
The EP80579 will consume 11 to 21 watts, depending on clock speed, and run between 600MHz and 1.2GHz. To satisfy industrial requirements, the company has announced a seven-year support life cycle.
Another chip, the EP80579, is for the embedded market. The next generation will be based on Atom instead of Pentium M, hardly a problem since Atom is also based on the Intel Architecture (IA) instruction set. The next generation of the MID processor line will be Lincroft, due next year, featuring a tenfold reduction in power consumption over the current line of chips.
For consumer electronics (CE), the company plans to release a chip code-named Canmore later this year, followed by Sodaville next year. Intel did not go into detail beyond saying that the chips would be tuned for consumer electronics devices, such as bringing the Internet to the television.
IDC analyst Shane Rau said this is a big change for Intel. "These are very highly integrated processors, he told InternetNews.com. "Intel is known for stand-alone processors," he said. "What's different about these is there's a system on a chip. So Intel is prepared to introduce pieces of silicon that are app-specific."
This reflects Intel's desire to look for every way possible to expand, and the way it expands is by finding new markets to enter. "It's the PC market looking to redeploy its technologies into new markets,
Intel Reveals First Details on Its GPU Entry
Intel has taken the wraps off its entry into the graphics processor market, offering up the first technical hints of "Larrabee," its attempt to take on nVidia and ATI/AMD in the highly competitive graphics processor space.
In a briefing with journalists ahead of its presentation at the Siggraph show later this month in Los Angeles, three engineers on the team went into great technical detail on the structure of the chip, but declined to give much product specification.
What they would say is that initially, Larrabee will be available as an add-in card, just like nVidia (NASDAQ: NVDA) and ATI (NYSE: AMD) cards are now. Other potential uses or the form factors of the cards were not discussed. Product won't appear on the market until late 2009, if not 2010.
Larrabee is built on the old Pentium technology, but heavily modified and modernized for graphics processing. Intel (NASDAQ: INTC) would not say how many cores would constitute a Larrabee processor, beyond the nebulous promise of "dozens." The cores all communicate through a wide "ring" bus that allows for fast inter-core communication and sharing of data, as well as sharing cache data. The L2 cache is partitioned among the cores, allowing for data replication and sharing.
Each Larrabee core is a complete x86 core capable of context switching and preemptive multitasking with support for virtual memory and page swapping. The main difference between Larrabee and other GPUs is that Larrabee will be much more flexible in the steps through what graphical data is processed.
Existing GPU architectures require data to be passed through a battery of processors, from a vertex shader to a pixel shader to a rasterizer, even if a particular processing job doesn't require it. As such, Intel believes that there is no such thing as a "typical" workload.
"The problem with designing a GPU is how much performance to put into the different segments to balance out variations in the load," said Larry Seiler, senior principal engineer in the visual computing group at Intel.
So Intel's solution is to not have fixed function stages in the pipeline. "You are entirely in control of how processing happens," said Seiler. "You can change scheduling, how each stage is handled, you can modify it to handle the characteristics of your workload, or change the rasterizer for your workload."
However, Jon Peddie, president of Jon Peddie Research, said Intel's solution is not necessarily better nor worse. "There are fixed functions and logical steps one has to go through in a GPU, but those things are in there for a very good reason. Those are the most efficient ways to do graphics programming," he told InternetNews.com.
What Intel is doing is adapting its strategy, the x86 architecture, to graphics, Peddie added. "This is really a multi core CPU. What makes it different from the x86 we are using in our computers is this ring communication for interprocessor communications. That is one of the main differentiators between Larrabee and Nehalem."
That said, Peddie thinks the "ring" for inter-core communication is a big revolution. "I think it's fantastic that Intel has done this, because this is the first innovation in computer graphics architecture since the GPU was introduced almost ten years ago. So they get a lot of credit from me for being brave enough to do it," he said.
The ring "gives you a really fat communication path for every processor to talk to every other processor. That's something they have that neither ATI nor nVidia have," said Peddie. "nVidia and ATI have an order of magnitude more processors, but are built in groups or gangs and communicate from group to group. So processor 004 can't talk directly to processor 794."
Seiler said that the flexibility of Larrabee is not limited to the hardware, but the software as well. "If a developer finds something in the API that limits them, they can create their own," he said. "We want to insure developers the freedom to run in Larrabee as they need."
That also means possible forking as developers improvise their own fixes, the same problem that made Unix so incompatible after many years of proprietary fixes. Intel is aware of that. "We want to give them freedom but we are wary of the potential for splintering. So it's a balancing act," said Seiler
Intel has been heavily romancing major computer graphics experts at universities all over the world and all of the major game developers. The paper being presented at Siggraph, along with many Intel engineers, includes Stanford engineers as contributors as well as Mike Abrash, one of the best known gaming graphics programmers.
So Intel is making a full court charge on Larrabee, a big change from its less-than-stellar integrated graphics products. "Don't judge Larrabee by Intel's current graphics products," said Peddie. "[Intel CEO] Paul Otellini has taken the handcuffs off the guys at Intel who know how to do graphics. Not only has he taken the cuffs off he's given them the checkbook to get some staff and IP behind it. As a result, Intel is going to do it right."
In a briefing with journalists ahead of its presentation at the Siggraph show later this month in Los Angeles, three engineers on the team went into great technical detail on the structure of the chip, but declined to give much product specification.
What they would say is that initially, Larrabee will be available as an add-in card, just like nVidia (NASDAQ: NVDA) and ATI (NYSE: AMD) cards are now. Other potential uses or the form factors of the cards were not discussed. Product won't appear on the market until late 2009, if not 2010.
Larrabee is built on the old Pentium technology, but heavily modified and modernized for graphics processing. Intel (NASDAQ: INTC) would not say how many cores would constitute a Larrabee processor, beyond the nebulous promise of "dozens." The cores all communicate through a wide "ring" bus that allows for fast inter-core communication and sharing of data, as well as sharing cache data. The L2 cache is partitioned among the cores, allowing for data replication and sharing.
Each Larrabee core is a complete x86 core capable of context switching and preemptive multitasking with support for virtual memory and page swapping. The main difference between Larrabee and other GPUs is that Larrabee will be much more flexible in the steps through what graphical data is processed.
Existing GPU architectures require data to be passed through a battery of processors, from a vertex shader to a pixel shader to a rasterizer, even if a particular processing job doesn't require it. As such, Intel believes that there is no such thing as a "typical" workload.
"The problem with designing a GPU is how much performance to put into the different segments to balance out variations in the load," said Larry Seiler, senior principal engineer in the visual computing group at Intel.
So Intel's solution is to not have fixed function stages in the pipeline. "You are entirely in control of how processing happens," said Seiler. "You can change scheduling, how each stage is handled, you can modify it to handle the characteristics of your workload, or change the rasterizer for your workload."
However, Jon Peddie, president of Jon Peddie Research, said Intel's solution is not necessarily better nor worse. "There are fixed functions and logical steps one has to go through in a GPU, but those things are in there for a very good reason. Those are the most efficient ways to do graphics programming," he told InternetNews.com.
What Intel is doing is adapting its strategy, the x86 architecture, to graphics, Peddie added. "This is really a multi core CPU. What makes it different from the x86 we are using in our computers is this ring communication for interprocessor communications. That is one of the main differentiators between Larrabee and Nehalem."
That said, Peddie thinks the "ring" for inter-core communication is a big revolution. "I think it's fantastic that Intel has done this, because this is the first innovation in computer graphics architecture since the GPU was introduced almost ten years ago. So they get a lot of credit from me for being brave enough to do it," he said.
The ring "gives you a really fat communication path for every processor to talk to every other processor. That's something they have that neither ATI nor nVidia have," said Peddie. "nVidia and ATI have an order of magnitude more processors, but are built in groups or gangs and communicate from group to group. So processor 004 can't talk directly to processor 794."
Seiler said that the flexibility of Larrabee is not limited to the hardware, but the software as well. "If a developer finds something in the API that limits them, they can create their own," he said. "We want to insure developers the freedom to run in Larrabee as they need."
That also means possible forking as developers improvise their own fixes, the same problem that made Unix so incompatible after many years of proprietary fixes. Intel is aware of that. "We want to give them freedom but we are wary of the potential for splintering. So it's a balancing act," said Seiler
Intel has been heavily romancing major computer graphics experts at universities all over the world and all of the major game developers. The paper being presented at Siggraph, along with many Intel engineers, includes Stanford engineers as contributors as well as Mike Abrash, one of the best known gaming graphics programmers.
So Intel is making a full court charge on Larrabee, a big change from its less-than-stellar integrated graphics products. "Don't judge Larrabee by Intel's current graphics products," said Peddie. "[Intel CEO] Paul Otellini has taken the handcuffs off the guys at Intel who know how to do graphics. Not only has he taken the cuffs off he's given them the checkbook to get some staff and IP behind it. As a result, Intel is going to do it right."
Apple's Suit Aims to Eclipse PsyStar
Apple has confirmed it's suing Doral, Florida-based startup PsyStar for selling computers using unauthorized copies of Apple's Leopard operating system.
"We take it very serious when we believe people have stolen our intellectual property," Apple spokesperson Susan Lundgren told InternetNews.com. Beyond that, Lundgren said Apple (NASDAQ: AAPL) had no other comment to make on the suit other than to confirm reports that Apple had taken the legal action against PsyStar last week.
Calls to PsyStar yesterday were not returned. Apple's suit was filed in U.S. District Court for the Northern District of California. Among Apple's complaints, is that PsyStar's use of Apple's proprietary software and intellectual property has "harmed consumers by selling to them a poor product that is advertised and promoted in a manner that falsely and unfairly implies an affiliation with Apple."
Apple co-founder and CEO Steve Jobs has never been a fan of licensing. When he returned to the company as CEO in 1997, one of his first acts was to shut down a licensing program that allowed other companies to build Mac-compatible computer systems powered by Apple's operating system.
PsyStar started selling a computer called the OpenMac back in May that included a modified version of Apple's operating system. As of today, the PsyStar Web site is still operating and taking orders. It touts the Open Computer as "The Smart Alternative to an Apple."
An FAQ on the site further states: "The idea of the Open Computers is not to pirate the Apple operating system but to allow the Apple operating system to be run on hardware of the user's choosing."
Among other products, PsyStar's cheapest is the Open Computer, a tower system with 2GB of memory, sans monitor, keyboard or mouse, that does include OS X Leopard pre-installed for $554.99.
The accompany product blurb says: "Why spend $1999 to get the least expensive Apple computer with a decent video card when you can pay less than a fourth of that for an equivalent sleek and small form-factor desktop with the same hardware."
Apple's cheapest Mac is the MacMini at $599, with less memory and graphics capability than the Open Computer.
PsyStar also sells downloaded updates for Leopard and servers using the operating system. Apple has asked the court to stop PsyStar from using Leopard and seeks unspecified damages, according to a report by the Bloomberg News service and numerous other news outlets.
"We take it very serious when we believe people have stolen our intellectual property," Apple spokesperson Susan Lundgren told InternetNews.com. Beyond that, Lundgren said Apple (NASDAQ: AAPL) had no other comment to make on the suit other than to confirm reports that Apple had taken the legal action against PsyStar last week.
Calls to PsyStar yesterday were not returned. Apple's suit was filed in U.S. District Court for the Northern District of California. Among Apple's complaints, is that PsyStar's use of Apple's proprietary software and intellectual property has "harmed consumers by selling to them a poor product that is advertised and promoted in a manner that falsely and unfairly implies an affiliation with Apple."
Apple co-founder and CEO Steve Jobs has never been a fan of licensing. When he returned to the company as CEO in 1997, one of his first acts was to shut down a licensing program that allowed other companies to build Mac-compatible computer systems powered by Apple's operating system.
PsyStar started selling a computer called the OpenMac back in May that included a modified version of Apple's operating system. As of today, the PsyStar Web site is still operating and taking orders. It touts the Open Computer as "The Smart Alternative to an Apple."
An FAQ on the site further states: "The idea of the Open Computers is not to pirate the Apple operating system but to allow the Apple operating system to be run on hardware of the user's choosing."
Among other products, PsyStar's cheapest is the Open Computer, a tower system with 2GB of memory, sans monitor, keyboard or mouse, that does include OS X Leopard pre-installed for $554.99.
The accompany product blurb says: "Why spend $1999 to get the least expensive Apple computer with a decent video card when you can pay less than a fourth of that for an equivalent sleek and small form-factor desktop with the same hardware."
Apple's cheapest Mac is the MacMini at $599, with less memory and graphics capability than the Open Computer.
PsyStar also sells downloaded updates for Leopard and servers using the operating system. Apple has asked the court to stop PsyStar from using Leopard and seeks unspecified damages, according to a report by the Bloomberg News service and numerous other news outlets.
Subscribe to:
Posts (Atom)
