Hasta La Gadget!

Massive supercomputers that devour electricity to keep them humming are not exactly the poster children for green technology. But IBM hopes to change that with its plans to build a supercomputer that will use water to keep the system cool and even recycle some of the waste heat to help heat the university where it’s housed.

The technology could lead to a reduction in overall energy consumption by at least 40 percent, when compared to similar air-cooled machines, says the company.

“Energy is arguably the number one challenge humanity will be facing in the 21st century,” says Dimos Poulikakos, lead investigator of the project. “We cannot afford anymore to design computer systems based on the criterion of computational speed and performance alone.”

Supercomputers are used in defense research labs such as Argonne National Laboratory, in space research by NASA and at universities for scientific research, all applications which have a nearly insatiable demand for processing power. The new supercomputer, called Aquasar, will be housed at the Swiss Federal Institute of Technology (ETH) Zurich and will have a top speed of 10 teraflops. (A teraflop is a trillion floating point operations per second, a measure of computing capacity.) While that’s a lot of computing power — a Core 2 Duo processor is capable of about 20 gigaflops, or 1/500 the speed of Aquasar — it’s a fraction of what some of the fastest supercomputers today. For instance, IBM’s Blue Gene/L supercomputer, which ranks fourth on the top 100 list, has a peak speed of 596 teraflops. Meanwhile, IBM has moved on to create its first supercomputer in Europe capable of one petaflop, or one thousand trillion operations per second.

Keeping these massive machines running isn’t as much a challenge as trying to maintain them in an optimal temperature band. Aquasar, however, hopes to offer more bang for the buck in terms of its energy consumption. Many of the chips used the supercomputing systems dissipate about ten times as much heat as a typical kitchen hotplate, says Thomas Brunschwiler, a researcher at IBM Zurich Research Lab. For optimal performance, the chips must be cooled below 185 degrees Fahrenheit (85 degrees Celsius).

Accomplishing that much cooling across a huge data center means a significant strain on electricity consumption. Researchers estimate that about 50 percent of an average air-cooled data center’s energy consumption stems from powering the cooling systems to keep the processors from overheating. Reducing that would be a big step towards energy efficiency.

The power consumption of one rack of the Aquasar will be around 10 KW, IBM officials say. By comparison, the Blue Gene L/P supercomputer consumes about 40 KW of power per rack, and the average power consumption of a supercomputer in the top 500 list is 257 KW.  Aquasar, set to be commissioned in 2010, will have two IBM BladeCenter servers in each rack.

Aquasar’s breakthrough lies in how it has successfully managed chip level water cooling, says Brunschwiler.

“One way to do it is to cool the air in a data center to 40 degrees Celsius (104 degrees Fahrenheit) , which means air conditioning units that take space and energy,” he says. “Or you can use liquid cooling to get there.”

In the Aquasar system, high performance micro-channel coolers are attached directly to the backside of the processor. In them, the cooler water is distributed through a fine network of capillaries that spread throughout the back.

It’s different from the water-cooled modules used in other supercomputers, says Brunschwiler. Water cooling on a module level brings the liquid between the processors, but not right up against them via micro capillaries.

“The breakthrough in our special package design lies in how we can bring the water as close as possible to the chips without letting it affect the chips’ performance,” says Brunschwiler.

The water-cooled supercomputer will require a small amount, just about 2.64 gallons of water for cooling. A pump ensures the water flows through at the rate of roughly 7.9 gallons per minute.

For overall efficiency, the entire cooling system is a closed circuit. The heated water from the chips is cooled as it passes through a passive heat exchanger and the removed heat is recycled. In this case, it is channeled into the University’s heating system.

“Heat is a valuable commodity that we rely on in our everyday lives,” says Bruno Michel, manager at IBM’s Zurich Research Laboratory. “If we capture and transport the waste heat from the active components in a computer system as efficiently as possible, we can reuse it as a resource.”

Photo: Aquasar/IBM Research

Research into color-changing nanoparticles could pave the way for a new kind of display technology. A breakthrough promises tiny molecules that can change color in response to an external magnetic field that can be used to create outdoor displays and posters.

“We have developed a new way to induce color change in materials that can be fabricated on a large scale and is pretty close to commercialization,” says Yadong Yin, an assistant professor of chemistry at University of California, Riverside, who led the study that included contributions from South Korean scientists.

The technique centers on polymer beads, called magnetochromatic microspheres, which are dispersed in a liquid such as water, alcohol or hexane.

Inside the beads are magnetic iron oxide nanostructures. Changing the orientation of the nanostructures with an external magnetic field helps produce the change in color of the beads.

The process is similar to the way electrophoretic displays, more commonly known as electronic ink, work. The two systems share common properties such as being bistable (stable in two distinct states), being readable in direct sunlight and consuming very little power.

To fabricate the polymer beads or microspheres, researchers mixed magnetic iron oxide particles into a resin. The resin solution was then dispersed in either mineral oil or silicon oil, which transformed the resin into spherical droplets in the oil. An external magnetic field organizes the iron oxide particles into periodically ordered chains that display a reflective color if viewed along the direction of the magnetic field.

“For instance, in a vertical field, the particle chains stand straight so that their diffraction is turned ‘on’ and and corresponding color can be observed from
the top,” say the researchers in their study.  When the field is switched horizontally, the microspheres are forced to rotate 90 degrees to lay down the particle chains so that the diffraction is turned off. The microspheres then
show the native brown color of iron oxide.  Depending on the direction of the external magnetic field there can also be intermediate stages.

As the final step, the liquid system which holds the particles is exposed to ultraviolet radiation to polymerize the resin droplets and make them into solid microspheres. This allows for switching between two states. The solid state allows for the color information to be frozen and retained for long times without the  need for additional power.

Yin did not explain exactly how many colors can be obtained from the display but said the system can handle a reasonably wide range, though switching to colors at the opposite ends of the spectrum could be a challenge.

The researchers published the result of their study in the latest issue of the Journal of the American Chemical Society.

Yin sees applications such large outdoor displays that can be expensive to do with LCDs or other display technologies. “If you want a huge LCD display outside the house it can be uneconomical,” he says. “We can do it for much cheaper with this new technology.”

The displays are reflective, so they can offer high visibility even in strong sunshine, says Yin. The new material also can be used to make environmentally friendly pigments for paints and cosmetics.

Here’s a quick video that shows the rotation of the microspheres in a vertically changing external magnetic field. The color is switched between on (blue) and off states.

Photo: Colorful microspheres/University of California, Riverside

A diamond is forever. And in a few years, you could say the same about everything you say on Twitter. Researchers from the University of California at Berkeley have found a way to develop a carbon nanotube-based technique for storing data that could potentially last more than a billion years.

The goal, say the researchers, is to improve on what they see as the general trend for memory storage. As memory density increases, the lifespan of the storage has been decreasing, they say.  For instance, stone carvings are still largely readable after 3,800 years, while information written with individual atoms by scanning tunneling microscopes last just a few seconds at room temperature. Conventional digital memory technologies in use — such as hard disk drives and flash memory — have an estimated lifetime of only 10 to 30 years. If successful, a billion-year memory storage device could change that, enabling humans to store any data — from the digital version of an ancient manuscript to your latest tweet — from now until long after the Earth has been overrun by superintelligent, fusion-powered cyborg ants.

Here’s how it works. The device has an iron nanoparticle positioned inside a hollow carbon nanotube. Carbon nanotubes are molecular-scale tubes usually made of a carbon allotrope. For data storage, a small electrical signal is applied across the nanotube causing the iron nanoparticle shuttle to move back and forth. The movement of the nanoparticles from one end to the other of the tube creates the binary ‘1′ or ‘0′ state.

The position of the shuttle can be read out directly, explain the researchers in a paper published in the current issue of the Nano Letters journal. “The reversibility of the nanoparticle motion allows a memory bit that can be rewritten,” according to the paper.

The technique has significant potential for archival storage, say the researchers, because the nanoparticle-based bits show significant persistence. It’s also possible to store a lot of data in a small space: With information density predicted to be as high as 10¹² bits per square inch, you could store data from nearly 25 DVDs in the space of a postage stamp.

The beauty of the system is that it requires only a couple volts of electrical signal to stimulate it, Will Gannett, a graduate student in physics working on the project at UC Berkeley told campus paper The Daily Californian.

It’ll take awhile to get there, though — so far the researchers have only demonstrated the theoretical possibility of this technology.

[via Science]

Photo: Nanoparticle in nanotube representation

Laptops run hot. Fast processors and skinny enclosures mean you have to shift a lot of air to keep things cool, something which requires noisy, battery-sucking fans. A new product from chip packaging company Tessera promises to help.

The Aeolian idea is to use as artificial, ionic wind to do the work. A small voltage-converter ups the potential of the notebook’s battery to 30,000 volts and this is fired between two electrodes. The result is an ionization of common molecules (nitrogen) which carry the rest of the air along with them, causing a brisk breeze to flow through the machine.

The advantages are many: The ion wind is silent, as there are no moving parts but the kiss of cool air, and the energy required is much lower — as little as half that of a regular fan system. The ion breeze also pulls out around 30% more heat.

The ionic breeze isn’t quite ready to blow into town, though. Problems with the electrodes corroding too quickly have not quite been solved, but Tessera has some competition in the form of a startup named Ventiva, which should hurry things along.

Thermal Management [Tessera via MIT via Oh Gizmo]

Image credit: Tessera

It can’t arm wrestle yet, but a robotic hand developed by students at Virginia Tech is strong enough to lift a can of food and dexterous enough to handle a raw egg.

It’s a big step for robotic hands, which have so far been hampered by lack of flexibility, forcing them to merely grab objects instead of being able to handle a wide range of textures and motions.

The latest hand, called RAPHaEL (Robotic Air Powered Hand with Elastic Ligaments), is powered by a compressed air tank. A microcontroller helps coordinate the motion of the fingers. The mechanism makes the hand deft enough to gesture for sign language.

The robotic hand’s grip depends on the extent of pressure of the air. A low pressure is used for a lighter grip, while a higher pressure allows for a sturdier grip.

“This air-powered design is what makes the hand unique as it does not require the use of any motors or other actuators,” said Dennis Hong, director and the faculty adviser on the project at the Robotics and Mechanisms Laboratory of Virgina Tech. “The grasping force and compliance can be easily adjusted by simply changing the air pressure.”

The hand could potentially be used to create robotic prosthetics, though at Virginia Tech it is part of a larger project. The university’s Robotics Lab is working to create a humanoid robot known as CHARLI (Cognitive Humanoid Robot with Learning Intelligence) that will be 5 feet tall and used as a research platform and in robot sports.

The latest version of the robotic hand is expected to be used in the CHARLI robot. Once the newer model hand is connected to the larger body, it will be able to pick up — not just grasp and hold — objects just like a person, says the lab.

Check out this video showing  RAPHaEL at work:

Photo: Robotic Hand/Virginia Tech College of Engineering