Pages

Showing posts with label IBM. Show all posts
Showing posts with label IBM. Show all posts

Thursday, 9 July 2015

IBM and allies find a way to make chips even tinier and faster

IBM Research and business partners have built a working chip with features measuring 7 nanometers, or billionths of a meter. That's half the size of today's cutting-edge 14nm chips.IBM Research
Chip designs two generations more advanced than today's cutting-edge designs are now closer to reality as IBM announced Wednesday it's built a test processor that makes computer circuitry significantly more powerful.
The test chip has working components, called transistors, but it is a research and development project rather than a finished product that can be built into a computing device like a laptop, server or smartphone. Nevertheless, it's an important step extending Moore's Law and its promise of steady progress in the computer industry.
The processor progress charted by Moore's Law has shrunk computers from refrigerator-sized hulks to smartphones that fit in your pocket. But it's getting harder to develop each new generation of chip technology, requiring years of materials research and manufacturing facilities costing in the vicinity of $10 billion. IBM's work signals that it'll be feasible to miniaturize chips further, helping to enable devices like powerful smartwatches or perhaps augmented-reality contact lenses.
"This is a welcome sign for the chip industry," said Envisioneering analyst Richard Doherty. "You can count on at least two more turns of Moore's Law benefits."
IBM Research led development of the technology with allies including electronics giant Samsung and chip manufacturer GlobalFoundries at a State University of New York nanoscale engineering project in Albany, New York. The work is part of the Common Platform alliance designed to speed research and the transition to new manufacturing methods. As the costs of developing and building next-gen chips rise, such alliances let companies pool their resources to better keep up with industry leader Intel.Moore's Law is named after Intel co-founder Gordon Moore, who 50 years ago noticed steady improvements in the number of transistors on a chip. Under Moore's Law, that number doubles every two years, unlocking new computing power and making it economical to squeeze processors into ever-smaller devices.
Even with allies, the work doesn't come cheap. IBM last year pledged to spend $3 billion over five years on research to continue diminishing the scale of chip features.

Seven nanometers or bust

Today's cutting-edge chips from Samsung and Intel are built with circuitry features measuring 14 nanometers, or 14 billionths of a meter. That's extraordinarily small: 14nm is 7,000 times narrower than a human hair, or alternately, six times wider than a strand of DNA.
One generation out will be chips with 10nm features that will double the circuitry for a given area. Two generations out come 7nm chips, and that's what IBM Research has demonstrated. For comparison, 7nm is less than three times the width of that 2.5nm DNA strand.
This comblike pattern on IBM Research's test chip shows the protruding "fins" in the foundational chip circuit elements called transistors.IBM Research
"It's a major step," said Mukesh Khare, vice president of semiconductor technology at IBM Research. "We have been working on this technology for more than five years."
IBM's 10nm technology improved the power-performance ratio by 40 percent or 50 percent over today's 14nm chips, meaning that a computer designer could either lower power consumption for better battery life or speed up software running on a computer. The 7nm design increases the power-performance ratio another 50 percent over the 10nm generation, Khare said.
IBM previously built its own chips, mostly for powerful servers that it sells to big businesses wanting track global inventory levels, find patterns in sales trends or host large-scale online services. In July, though, GlobalFoundries announced itcompleted its acquisition of IBM's microelectronics business; IBM will pay GlobalFoundries to build its chips for the next 10 years.
Intel has led the industry in the development of new manufacturing processes, introducing a new generational "shrink" every two years. It's not clear yet how Intel will move to 10nm and 7nm chips, though.
"Intel has said almost nothing publicly that's concrete about 7nm development," said Forrester analyst Richard Fichera. "It's real hard to bet against Intel in the long run, but [IBM Research's 7nm work] clearly says they have somebody breathing down the back of their neck."

The road to 7nm

IBM Research and its allies employed a number of technologies to make its 7nm prototype real. Two big ones are a chemical compound called silicon germanium and an optical etching technology using extreme ultraviolet light.
Researchers hold a silicon wafer 300mm in diameter that houses dozens of test chips.
Researchers hold a silicon wafer 300mm in diameter that houses dozens of test chips.IBM Research
Computer chips are built on a disk-shaped substrate of silicon crystal called a wafer, but chipmakers have long fiddled with the wafer's exact chemical composition. Doing so can mean better electrical properties for transistors, the tiny on-off switches that in their millions or billions make up a modern microchip. For IBM Research's 7nm chips, adding a layer of silicon germanium makes the transistors switch on and off faster, Khare said. The chip can therefore process data faster, meaning there's less of a delay to apply that Instagram photo filter or to draw that Starcraft spaceship on the screen.
The extreme ultraviolet light is used for the etching of circuitry patterns on the silicon wafer -- a fundamental part of chip manufacturing. This etching process, called photolithography, shines light through a mask that has an extremely complicated arrangement of transparent and opaque areas. Where the light shines or doesn't changes the composition of the wafer, which means different types of materials can be added removed to manufacture the three-dimensional transistors and interconnection circuitry.

Finger-painting with a boxing glove

Photolithography patterns have shrunk along with chip circuitry sizes, but for the last decade or so, chipmakers have used invisible ultraviolet light with a wavelength of 193nm. That's remarkable, given that it's something like detailed finger-painting with a boxing glove. But by using a succession of two or three specially created masks for each layer of a chip pattern, chipmakers can construct very small-scale features.
For the 7nm chips, IBM Research uses extreme ultraviolet light, with a wavelength of 13.5nm that permits much smaller features.
"It's really hard," Doherty said of the extreme ultraviolet (EUV) transition. "These invisible light waves are almost X-rays in wavelength! The optics are different, the masks, the materials -- everything."
The chip industry has been expecting extreme ultraviolet for years, though, and the transition pain will pay off with a lithography process that can support future chipmaking generations, too, Khare said.
"Scaling of semiconductor technology is getting harder and harder," Khare said. "The business-as-usual conventional techniques do not apply."

Tuesday, 31 March 2015

IBM to invest $3B in building Internet of Things business

109503852.jpg
IBM is getting serious about the so-called Internet of Things, the idea of connecting more devices and objects to the Web.Sean Gallup/Getty Images
IBM on Tuesday said that it will create a new Internet of Things unit and invest $3 billion over four years to build it out.
The move formalizes IBM's existing Internet of Things efforts. IBM's smarter-planet and smarter-cities businesses are connected to the Internet of Things trend. The rough idea behind the Internet of Things is that sensors will be embedded in everything and networked to create data. This flow of data could improve operations.
For IBM, the formation of the Internet of Things unit follows a familiar playbook. IBM targets a high-value growth area, invests at least a $1 billion to get the effort rolling and throws its hardware, software and consultants at the issue. In this respect, the formation of the Internet of Things unit rhymes with what IBM did with e-commerce, analytics, and cloud and cognitive computing.
IBM faces a fierce battle for enterprise Internet of Things (IoT) business. Cisco has targeted IoT, as has almost every tech vendor.
Meanwhile, nontraditional IBM rivals have strong IoT efforts. For instance, General Electric, which happens to make many of the things that will be networked, has an IoT platform called Predix. GE has invested $1 billion in industrial software development. Although GE calls the Internet of Things the industrial Internet, the concept of networking things and layering analytics on top is the same.
For IBM's part, the company said it will have more than 2,000 consultants, researchers and developers aimed at IoT and the analytics that go with it. IBM said the unit will include:
  • A cloud platform for industries aimed at verticals. IBM will offer dynamic pricing models and cloud delivery to various verticals.
  • Bluemix IoT platform as a service so developers can create and deploy applications for asset tracking, facilities management and engineering tools.
    • An ecosystem of partners ranging from AT&T to ARM to The Weather Company.
    Separately, IBM announced a partnership with the business-to-business division of The Weather Company, owner of The Weather Channel. The partnership will deliver micro weather forecasts using sensors from aircraft, drones, buildings and smartphones.
    The Weather Company will also move its data services platform to IBM's cloud platform and integrate Big Blue's analytics tools such as Watson Analytics.
    To be sure, IBM has a bevy of IoT projects under way with customers. The new unit will hone and focus those efforts while bringing in IBM's expertise in analytics.

Friday, 18 October 2013

How IBM is making computers more like your brain. For real


IBM Research is working on "interlayer cooling," in which water is pumped through tiny tubes penetrating chips are piggypacked using high-speed communication technology called through-silicon vias. IBM's approach is designed to deal with overheating problems that otherwise severely limit chip stacking. The protruding pipe fittings are for connecting water-cooling tubes.
(Credit: Stephen Shankland/CNET)

ZURICH, Switzerland -- Despite a strong philosophical connection, computers and brains inhabit separate realms in research. IBM, though, believes the time is ripe to bring them together.
Through research projects expected to take a decade, Big Blue is using biological and manufactured forms of computing to learn about the other.
On the computing side, IBM is using the brain as a template for breakthrough designs such as the idea of using fluids both to cool the machine and to distribute electrical power. That could enable processing power that's densely packed into 3D volumes rather than spread out across flat 2D circuit boards with slow communication links.
And on the brain side, IBM is supplying computing equipment to a $1.3 billion European effort called the Human Brain Project. It uses computers to simulate the actual workings of an entire brain -- a mouse's first, then a human's -- all the way down to the biochemical level of the neuron. Researchers will be able to tweak parameters as the simulation is running to try to figure out core mechanisms for conditions like Alzheimer's disease, schizophrenia, and autism.

IBM's brain-inspired computing technology (pictures)

1-2 of 10
Scroll LeftScroll Right
It's all part of what IBM calls the cognitive systems era, in which computers aren't just programmed, but also perceive what's going on, make judgments, communicate with natural language, and learn from experience. It's a close cousin to that decades-old dream of artificial intelligence.
"If we want to make an impact in the cognitive systems era, we need to understand how the brain works," said Matthias Kaiserswerth, a computer scientist who's director of IBM Research in Zurich, speaking during a media tour of the labs on Wednesday.
One key challenge driving IBM's work is matching the brain's power consumption. Over millions of years, nature has evolved a remarkably efficient information-processing design, saidAlessandro Curioni, manager of IBM Research's computational sciences department. The ability to process the subtleties of human language helped IBM's Watson supercomputer win at "Jeopardy." That was a high-profile step on the road to cognitive computing, but from a practical perspective, it also showed how much farther computing has to go.
"Watson used 85 kilowatts," Kaiserwerth said. "That's a lot of power. The human brain uses 20 watts."
Bruno Michel describes Aquasar, an IBM Research prototype high-performance computing machine that uses unusually high-temperature liquid cooling.
Bruno Michel describes Aquasar, an IBM Research prototype high-performance computing machine that uses unusually high-temperature liquid cooling.
(Credit: Stephen Shankland/CNET)
Dense 3D computing
The shift in IBM's computing research shows in the units the company uses to measure progress. For decades, the yardstick of choice for gauging computer performance has been operations per second -- the rate at which the machine can perform mathematical calculations, for example.
When energy constraints became a problem, meaning that computers required prohibitive amounts of electrical power and threw off problematic amounts of waste heat, a new measurement arrived: operations per joule of energy. That gauges a computer's energy efficiency.
Now IBM has a new yardstick: operations per liter. The company is judging success by how much data-processing ability it can squeeze into a given volume. Today's computers must be laid out on flat circuit boards that ensure plenty of contact with air that cools the chips.
"In a computer, processors occupy one-millionth of the volume. In a brain, it's 40 percent. Our brain is a volumetric, dense, object," said Bruno Michel, a researcher in advanced thermal packaging for IBM Research, who got his Ph.D in biophysics.
What's the problem with sprawl? In short, communication links between processing elements can't keep up with data-transfer demands, and they consume too much power as well, Michel said.
The fix is to stack chips into dense 3D configurations, with chips linked using a technology called through-silicon vias (TSVs). That's impossible today because stacking even two chips means crippling overheating problems. But IBM believes it's got an answer to the cooling problem: a branching network of liquid cooling channels that funnel fluid into ever-smaller tubes.
The liquid passes not next to the chip, but through it, drawing away heat in the thousandth of a second it takes to make the trip, Michel said. The company has demonstrated the approach in an efficient prototype system called Aquasar. (Get ready for another new yardstick: greenhouse gas emissions. Aquasar can perform 7.9 trillion operations per second per gram of carbon dioxide released into the atmosphere.)
IBM can deliver up to 1 watt of power per square centimeter with this technology called a flow battery, which transports electrical power stored chemically. Here, vanadium electrolytes power a microfluidics chip in a lab demonstration. Ultimately IBM hopes to use liquids both to cool and power computers.
IBM can deliver up to 1 watt of power per square centimeter with this technology called a flow battery, which transports electrical power stored chemically. Here, vanadium electrolytes power a microfluidics chip in a lab demonstration. Ultimately IBM hopes to use liquids both to cool and power computers.
(Credit: Stephen Shankland/CNET)
Liquid-based flow battery
But that's not all the liquid will do. IBM also is developing a system called a redox flow battery that also uses it to distribute power instead of using wires. Two liquids called electrolytes, each with oppositely charged electrical ions, circulate through the system to distribute power. Think of it as a liquid battery interlaced through the interstices of the machine.
"We are going to provide cooling and power with a fluid," Michel said. "That's how our brain does it."
The electrolytes, vanadium-based at present, travel through ever-smaller tubes, said Patrick Ruch, another IBM researcher working on the effort. At the smallest, they're about 100 microns wide, about the width of a human hair, at which point they hand off their power to conventional electrical wires. Flow batteries can produce between 0.5 and 3 volts, and that in turn means IBM can use the technology today to supply 1 watt of power for every square centimeter of a computer's circuit board.
Liquid cooling has been around for decades in the computing industry, but most data centers avoid it given its expense and complexity. It's possible the redox battery could provide a new incentive to embrace it, though.
Michel estimates the liquid power technology will take 10 to 15 years to develop, but when it works, it'll mean supercomputers that fit into something the size of a backpack, not a basketball court.
"A 1-petaflop computer in 10 liters -- that's our goal," Michel said.
Performing at 1 petaflop means a computer can complete a quadrillion floating-point mathematical operations per second. Today's top supercomputer clocked in at 33.86 petaflops, but it uses 32,000 Xeon processors and 48,000 Xeon Phi accelerator processors.
Matthias Kaiserswerth, director of IBM Research in Zurich, is working toward the era of "cognitive computing," in which machines get attributes of human thinking such as perception, learning, and judgment.
Matthias Kaiserswerth, director of IBM Research in Zurich, is working toward the era of "cognitive computing," in which machines get attributes of human thinking such as perception, learning, and judgment.
(Credit: Stephen Shankland/CNET)
How to build a brain
More conventional supercomputers have been used so far for IBM's collaborations in brain research. The highlight of that work so far has been the Blue Brain project, which is on its thirdIBM Blue Gene supercomputer at the Ecole Polytechnique Federale de Lausanne, or EPFL, in Lausanne, Switzerland. The Blue Brain and Human Brain Project will take a new step with aBlue Gene/Q augmented by 128 terabytes of flash memory at the Swiss National Supercomputing Center in Lugano, Switzerland. It'll be used to simulate the formation and inner workings of an entire mouse brain, which has about 70 million neurons.
The eventual human brain simulation will take place at the Juelich Supercomputing Center in northern Germany, Curioni said. It's planned to be an "exascale" machine -- one that performs 1 exaflops, or quintillion floating-point operations per second.
The project doesn't lack for ambition. One of its driving forces is co-director Henry Markram of EPFL, who has worked on the Blue Brain project for years and sees computing as the way to understand the true workings of the human brain.
"It's impossible to experimentally map the brain," simply because it's too complicated, Markram said. There are too many neurons overall, 55 different varieties of neuron, and 3,000 ways they can interconnect. That complexity is multiplied by differences that appear with 600 different diseases, genetic variation from one person to the next, and changes that go along with the age and sex of humans.
"If you can't experimentally map the brain, you have to predict it -- the numbers of neurons, the types, where the proteins are located, how they'll interact," Markram said. "We have to develop an entirely new science where we predict most of the stuff that cannot be measured."
Liquid cooling has traditionally meant water traveling near chips, the hottest part of computers, but IBM Research has begun making chips with cooling conduits built directly in.
Liquid cooling has traditionally meant water traveling near chips, the hottest part of computers, but IBM Research has begun making chips with cooling conduits built directly in.
(Credit: Stephen Shankland/CNET)
With the Human Brain Project, researchers will use supercomputers to reproduce how brains form -- basically, growing them in an virtual vat -- then seeing how they respond to input signals from simulated senses and nervous system.
The idea isn't to reproduce every last thing about the brain, but rather a model based on the understanding so far. If it works, actual brain behavior should emerge from the fundamental framework inside the computer, and where it doesn't work, scientists will know where their knowledge falls short.
"We take these rules and algorithmically reconstruct a model of the brain," Markram said. "We'll say this is biological prediction, then we can go back to the experiments and we can verify if the model is right. We celebrate when the model is wrong, because that's when it points to where we need more data or we don't understand the rules."
The result, if the work is successful, will be not just a better understanding of the brain, but better cooperation among brain researchers and medical experts. That could reverse recent declines in the development of new drugs to treat neural problems, he said.
And understanding the brain could usher in the era of "neuromorphic computing."
"Any new rules, circuits, or understanding of how the brain works will allow us to design neuromorphic machines that are much more powerful in terms of cognitive power, energy efficiency, and packaging," Curioni said.
And that, in turn, could lead to profoundly more capable computers. For starters, IBM has four markets in mind: machines that could find the best places to invest money, bring new depth and accuracy to medical diagnoses, research the appropriate legal precedents in court cases, or give people help when they dial a call center.
But it's not hard to imagine that's only the beginning. When computers can learn for themselves and program themselves, it's clear the divide separating biological and artificial computing will be a lot narrower.
IBM Research investigates supercomputing, nanotechnology, medicine, and more at its Zurich labs.
IBM Research investigates supercomputing, nanotechnology, medicine, and more at its Zurich labs.
(Credit: Stephen Shankland/CNET)