At the recent fall Intel Developer Forum in San Francisco, president and CEO, Paul Otellini announced shipment dates for the company’s latest quad-core processors and outlined its future strategy for 45 nanometre and 32 nanometre technology. These developments, he stated, ensured that Intel had now regained its leadership role in the microprocessor ‘war’ its been fighting against arch-rival AMD. Mark Davison reports.
If anyone thought that current processor power was sufficient for all of today’s computing needs, and that Moore’s Law was about to be tossed out the window, think again. According to Intel president and CEO, Paul Otellini, we ain’t seen nothin’ yet.
Otellini opened his keynote at this year’s fall IDF in San Francisco with a brief wrap of developments over the past few months.
“Computers are like people,” he began, “despite the differences on the outside, it’s what’s inside that counts. Not long ago – in 2001 – people were saying we didn’t need any more processing power – the only ones that needed extra were NASA scientists and designers.
“Last month, PC Magazine said that speed matters again – now, more than ever, performance matters again.
“So what’s changed?” Otellini asks. “Simply, we have … and what we do with our computers has changed. The things people want to do require more processing power than ever. Games, Google, finding data, photo editing, antivirus, OS10, Vista … these are the kinds of things that are driving the need for more computing power.
“Take YouTube, for example,” Otellini continues. “A 60 second video clip would have used 100% of the chip’s power in 2003; in 2004, this would have been down to around 40%; today, it’s just a few percent of microprocessor power.
“That leaves the capacity to do other things. But video won’t just stay where it is today. It will evolve. And high-definition (HD) this Christmas will reach a price point with cameras, for example, dropping to around $100. There is going to be huge uptake, but you need eight times the processing power for HD than you do for regular video.
“More than ever, processing power matters, even as the need to reduce heat, extend battery life and reduce electricity costs in data centres becomes more critical,” says Otellini. “Silicon technology is at the heart of the solution. It is how we get there.”
When it comes to performance and energy efficiency, Intel’s Core micro-architecture and Core Duo processor have set a new standard for the industry, Otellini says.
“Core Duo has been one of the most significant announcements we have ever made – it was a stunning leap ahead,” he says. “Demand has been strong and we have shipped five million units in the 60 days since its release.
“This is the fastest 60-day ramp for both desktop and notebook in Intel’s history.
“So what’s next?” he asks. “After dual-core comes quad-core and we’re going to accelerate this from Q1 of 2007 to Q4 of 2006. The initial products – Intel Core Extreme quad-core – will be targeted at enthusiasts and gamers and will ship in November.
“The Intel Core2 Extreme is currently the fastest microprocessor on the planet,” Otellini says. “Core Extreme quad-core will be 70% faster.”
The company’s mainstream quad-core processor will be shipped in the first quarter of 2007 and will be called the Intel Core Quad processor.
For servers, the Quad-Core Xeon processor 5300 series brand for dual processor servers will be shipped this year, and a new low-power 50-watt Quad-Core Xeon processor L5310 for blade servers will be shipped in the first quarter of 2007.
Performance and energy efficiency “all start with the transistor” Otellini says, describing Intel’s legacy of advancing Moore’s Law and its industry-leading silicon technology and manufacturing capability.
Intel, he adds, was the first to implement advanced 65nm silicon manufacturing technology in 2005, integrating power-saving features into the process that were critical to delivering power-efficiency at the transistor level.
Otellini says the company is now officially shipping the majority of its processors on 65nm. “This year, we will ship 40-million 65nm units. The rest of the industry? Zero.”
Looking ahead, Intel’s next-generation 45nm technology is on track for production in the second half of 2007 as planned, and Otellini disclosed for the first time that the company has 15 45nm products already in development across desktop, mobile and enterprise segments.
The first of these products is on track to complete its design in the fourth quarter of this year. He described the company’s extensive 45nm factory network with more than 500 000 square feet of clean room space and more than $9-billion invested.
Otellini estimates that the “cadence” of these manufacturing process technologies which follow Moore’s Law, coupled with Intel’s plans to introduce new micro-architectures about every two years, will result in significant performance-per-watt improvement over today’s Core micro-architecture products by 2010.
He showed a chart that mapped out new micro-architectures coming in 2008 (codenamed Nehalem and targeted at 45nm) followed by another in 2010 (codenamed Gesher and targeted at 32nm). These new micro architectures will be developed by separate teams working in parallel, and targeted for intersection with specific future process technologies.
“By the end of the decade we will deliver a 300% increase in performance per watt over today’s processors,” he says. “This improved power and performance will enable developers and manufacturers to develop systems with incredibly exciting new capabilities.”
To demonstrate how Moore’s Law will continue well into the future, Otellini showed a new research prototype processor that has 80 floating point cores on a single die.
The tiny silicon die on this experimental chip, just 300 square millimetres, is capable of achieving a Teraflop of performance, or 1-trillion floating point operations per second.
He contrasted this with Intel’s historic breakthrough 11 years ago with the world’s first Teraflop supercomputer, a massive machine powered by nearly 10 000 Pentium Pro processors in more than 85 large cabinets occupying about 2 000 square feet.
And the cost? A mere snip at $50-million.