Berkeley Lab has signed a collaboration agreement with Tensilica, Inc. to explore the use of Tensilica’s Xtensa processor cores as the basic building blocks in a massively parallel system design. Tensilica’s Xtensa processor is about 400 times more efficient in floating point operations per watt than the conventional server processor chip shown here.

In a paper published in the May issue of the International Journal of High Performance Computing Applications, Michael Wehner and Lenny Oliker of Berkeley Lab’s Computational Research Division, and John Shalf of the National Energy Research Scientific Computing Center (NERSC) lay out the benefit of a new class of supercomputers for modeling climate conditions and understanding climate change. Using the embedded microprocessor technology used in cell phones, iPods, toaster ovens and most other modern day electronic conveniences, they propose designing a cost-effective machine for running these models and improving climate predictions.

In April, Berkeley Lab signed a collaboration agreement with Tensilica, Inc. to explore such new design concepts for energy-efficient high-performance scientific computer systems. The joint effort is focused on novel processor and systems architectures using large numbers of small processor cores, connected together with optimized links, and tuned to the requirements of highly-parallel applications such as climate modeling.

Understanding how human activity is changing global climate is one of the great scientific challenges of our time. Scientists have tackled this issue by developing climate models that use the historical data of factors that shape the earth’s climate, such as rainfall, hurricanes, sea surface temperatures and carbon dioxide in the atmosphere. One of the greatest challenges in creating these models, however, is to develop accurate cloud simulations.

Although cloud systems have been included in climate models in the past, they lack the details that could improve the accuracy of climate predictions. Wehner, Oliker and Shalf set out to establish a practical estimate for building a supercomputer capable of creating climate models at 1-kilometer (km) scale. A cloud system model at the 1-km scale would provide rich details that are not available from existing models.

To develop a 1-km cloud model, scientists would need a supercomputer that is 1,000 times more powerful than what is available today, the researchers say. But building a supercomputer powerful enough to tackle this problem is a huge challenge.

Historically, supercomputer makers build larger and more powerful systems by increasing the number of conventional microprocessors — usually the same kinds of microprocessors used to build personal computers. Although feasible for building computers large enough to solve many scientific problems, using this approach to build a system capable of modeling clouds at a 1-km scale would cost about $1 billion. The system also would require 200 megawatts of electricity to operate, enough energy to power a small city of 100,000 residents.

In their paper, Towards Ultra-High Resolution models of Climate and Weather, the researchers present a radical alternative that would cost less to build and require less electricity to operate. They conclude that a supercomputer using about 20 million embedded microprocessors would deliver the results and cost $75 million to construct. This “climate computer” would consume less than 4 megawatts of power and achieve a peak performance of 200 petaflops.

“Without such a paradigm shift, power will ultimately limit the scale and performance of future supercomputing systems, and therefore fail to meet the demanding computational needs of important scientific challenges like the climate modeling,” Shalf said.

The researchers arrive at their findings by extrapolating performance data from the Community Atmospheric Model (CAM). CAM, developed at the National Center for Atmospheric Research in Boulder, Colorado, is a series of global atmosphere models commonly used by weather and climate researchers.

The “climate computer” is not merely a concept. Wehner, Oliker and Shalf, along with researchers from UC Berkeley, are working with scientists from Colorado State University to build a prototype system in order to run a new global atmospheric model developed at Colorado State.

“What we have demonstrated is that in the exascale computing regime, it makes more sense to target machine design for specific applications,” Wehner said. “It will be impractical from a cost and power perspective to build general-purpose machines like today’s supercomputers.”

Under the agreement with Tensilica, the team will use Tensilica’s Xtensa LX extensible processor cores as the basic building blocks in a massively parallel system design. Each processor will dissipate a few hundred milliwatts of power, yet deliver billions of floating point operations per second and be programmable using standard programming languages and tools. This equates to an order-of-magnitude improvement in floating point operations per watt, compared to conventional desktop and server processor chips. The small size and low power of these processors allows tight integration at the chip, board and rack level and scaling to millions of processors within a power budget of a few megawatts.

Berkeley Lab is a U.S. Department of Energy national laboratory located in Berkeley, California. It conducts unclassified scientific research and is managed by the University of California. Visit our Website at www.lbl.gov.

[Ucilia Wang @ DOE/Lawrence Berkeley National Laboratory]

Scene: The 2008 RALPH Awards, held at a highly secret location somewhere in the Northern Hemisphere. via Reno Gazette-Journal

A new game, named Foldit, turns protein folding into a competitive sport. Introductory levels teach the rules, which are the same laws of physics by which protein strands curl and twist into three-dimensional shapes - key for biological mysteries ranging from Alzheimer’s to vaccines.

After about 20 minutes of training, people feel like they’re playing a video game but are actually mouse-clicking in the name of medical science. The free program can be found here.

The game was developed by doctoral student Seth Cooper and postdoctoral researcher Adrien Treuille, both in computer science and engineering, working with Zoran Popovic, a UW associate professor of computer science and engineering; David Baker, a UW professor of biochemistry and Howard Hughes Medical Institute investigator; and David Salesin, a UW professor of computer science and engineering. Professional game designers provided advice during the game’s creation.

“We’re hopefully going to change the way science is done, and who it’s done by,” said Popovic, who presented the project today at the Games for Health meeting in Baltimore. “Our ultimate goal is to have ordinary people play the game and eventually be candidates for winning the Nobel Prize.”

Proteins, of which there are more than 100,000 different kinds in the human body, form every cell, make up the immune system and set the speed of chemical reactions. We know many proteins’ genetic sequence, but don’t know how they fold up into complex shapes whose nooks and crannies play crucial biological roles.

Computer simulators calculate all possible protein shapes, but this is a mathematical problem so huge that all the computers in the world would take centuries to solve it. In 2005, Baker developed a project named Rosetta@home that taps into volunteers’ computer time all around the world. But even 200,000 volunteers aren’t enough.

“There are too many possibilities for the computer to go through every possible one,” Baker said. “An approach like Rosetta@home does well on small proteins, but as the protein gets bigger and bigger it gets harder and harder, and the computers often fail.

“People, using their intuition, might be able to home in on the right answer much more quickly.”

Rosetta@home and Foldit both use the Rosetta protein-folding software. Foldit is the first protein-folding project that asks volunteers for something other than unused processor cycles on their computers or Playstation machines. Foldit also differs from recent human-computer interactive games that use humans’ ability to recognize images or interpret text. Instead, Foldit capitalizes on people’s natural 3-D problem-solving skills.

The intuitive skills that make someone good at playing Foldit are not necessarily the ones that make a top biologist. Baker says his 13-year-old son is faster at folding proteins than he is. Others may be even faster.

“I imagine that there’s a 12-year-old in Indonesia who can see all this in their head,” Baker says.

Eventually, the researchers hope to advance science by discovering protein-folding prodigies who have natural abilities to see proteins in 3-D.

“Some people are just able to look at the game and in less than two minutes, get to the top score,” said Popovic. “They can’t even explain what they’re doing, but somehow they’re able to do it.”

The game looks like a 21st-century version of Tetris, with multicolored geometric snakes filling the screen. A team that includes a half-dozen UW graduate and undergraduate students spent more than a year figuring out how to make the game both accurate and engaging. They faced some special challenges that commercial game developers don’t encounter.

“We don’t know what the best result is, so we can’t help people or hint people toward that goal,” Popovic explained. The team also couldn’t arbitrarily decide to make one move worth 1,000 bonus points, since the score corresponds to the energy needed to hold the protein in that shape.

Almost 1,000 players have tested the system in recent weeks, playing informal challenges using proteins with known shapes. Starting this week, however, the developers will open the game to the public and offer proteins of unknown shapes. Also starting this week, Foldit gamers will face off against research groups around the world in a major protein-structure competition held every two years.

Beginning in the fall, Foldit problems will expand to involve creating new proteins that we might wish existed - enzymes that could break up toxic waste, for example, or that would absorb carbon dioxide from the air. Computers alone cannot design a protein from scratch. The game lets the computer help out when it’s a simple optimization problem - the same way that computer solitaire sometimes moves the cards to clean up the table - letting the player concentrate on interesting moves.

Eventually, the researchers hope to present a medical nemesis, such as HIV or malaria, and challenge players to devise a protein with just the right shape to lock into the virus and deactivate it. Winning protein designs will be synthesized in Baker’s lab and tested in petri dishes. High-scoring players will be credited in scientific publications the way that top Rosetta@home contributors already are credited for their computer time.

“Long-term, I’m hoping that we can get a significant fraction of the world’s population engaged in solving critical problems in world health, and doing it collaboratively and successfully through the game,” Baker said. “We’re trying to use the brain power of people all around the world to advance biomedical research.”

Foldit includes elements of multiplayer games in which people can team up, chat with other players and create online profiles. Over time the researchers will analyze people’s moves to see how the top players solve puzzles. This information will be fed back into the game’s design so the game’s tools and format can evolve.

[Hannah Hickey @ University of Washington]

Engineers at Duke University believe that the results of feasibility studies conducted in their laboratory represent the first concrete steps toward achieving this space age vision of the future. Also, on a more immediate level, the technology developed by the engineers could make certain contemporary medical procedures safer for patients, they said.

For their experiments, the engineers started with a rudimentary tabletop robot whose “eyes” used a novel 3-D ultrasound technology developed in the Duke laboratories. An artificial intelligence program served as the robot’s “brain” by taking real-time 3-D information, processing it, and giving the robot specific commands to perform.

“In a number of tasks, the computer was able to direct the robot’s actions,” said Stephen Smith, director of the Duke University Ultrasound Transducer Group and senior member of the research team. “We believe that this is the first proof-of-concept for this approach. Given that we achieved these early results with a rudimentary robot and a basic artificial intelligence program, the technology will advance to the point where robots — without the guidance of the doctor — can someday operate on people.”

The results of a series of experiments on the robot system directing catheters inside synthetic blood vessels was published online in the journal IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control. A second study, published in April in the journal Ultrasonic Imaging, demonstrated that the autonomous robot system could successfully perform a simulated needle biopsy.

Advances in ultrasound technology have made these latest experiments possible, the researchers said, by generating detailed, 3-D moving images in real-time.

The Duke laboratory has a long track record of modifying traditional 2-D ultrasound — like that used to image babies in utero — into the more advanced 3-D scans. After inventing the technique in 1991, the team also has shown its utility in developing specialized catheters and endoscopes for real-time imaging of blood vessels in the heart and brain.

In the latest experiment, the robot successfully performed its main task: directing a needle on the end of the robotic arm to touch the tip of another needle within a blood vessel graft. The robot’s needle was guided by a tiny 3-D ultrasound transducer, the “wand” that collects the 3-D images, attached to a catheter commonly used in angioplasty procedures.

“The robot was able to accurately direct needle probes to target needles based on the information sent by the catheter transducer,” said John Whitman, a senior engineering student in Smith’s laboratory and first author on both papers. “The ability of the robot to guide a probe within a vascular graft is a first step toward further testing the system in animal models.”

While the research will continue to refine the ability of robots to perform independent procedures, the new technology could also have more direct and immediate applications.

“Currently, cardiologists doing catheter-based procedures use fluoroscopy, which employs radiation, to guide their actions,” Smith said. “Putting a 3-D ultrasound transducer on the end of the catheter could provide clearer images to the physician and greatly reduce the need for patients to be exposed to radiation.”

In the earlier experiments, the tabletop robot arm successfully touched a needle on the arm to another needle in a water bath. Then it performed a simulated biopsy of a cyst, fashioned out of a liquid-filled balloon in a medium designed to simulate tissue.

“These experiments demonstrated the feasibility of autonomous robots accomplishing simulated tasks under the guidance of 3-D ultrasound, and we believe that it warrants additional study,” Whitman said.

The researchers said that adding this 3-D capability to more powerful and sophisticated surgical robots already in use at many hospitals could hasten the development of autonomous robots that could perform complex procedures on humans.

[Richard Merritt @ Duke University]