‘Proximity Communication’ May Speed Drug Simulations


By Stacy Collett Computerworld

Medical researchers imagine a day when gene-mapping simulations will take an hour or two to compute instead of a week, on systems that process data up to 100 times faster than they do today. That dream could become a reality in the next five years with technology called proximity communication that's under development at Sun Microsystems Laboratories in Menlo Park, Calif.

For years, Sun has been looking for ways to "get computers back on the improvement curve they've fallen off of," says principal research scientist Robert Drost. While processor speeds have improved up to 4,000 percent, I/O speeds have improved by a factor of just 10 to 15. Proximity communication could catapult I/O speeds into the stratosphere.

In early 2000, research fellow and Sun Labs Vice President Ivan Sutherland hatched an idea while visiting his friend Steve Jacobson at Sarcos Research Corp., a Salt Lake City-based manufacturer of robotics and advanced prosthetics. Developers there were using a process called capacitive sensing to measure minute shifts in the movement between a chip and a glass plate -- movements as small as one millionth of a centimeter.

Sutherland hypothesized that if tiny mechanical movements could be accomplished through capacitive sensing, then surely it should be possible to put chips next to one another and send data using that same physical phenomenon.

Sutherland developed chip-to-chip communication technology and then enlisted the expertise of Drost, who at 34 was named to the 2004 list of the world's 100 Top Young Innovators Under Age 35 by MIT's Technology Review.

Proximity communication is a process where two chips, each with transmitter and receiver circuits, are positioned extremely close to each other. Without wires or soldered connections, data is transmitted across the gap by "capacitive coupling," which is coupling between charged particles that are at rest. Fewer wires and connections mean fewer bottlenecks in multiprocessor computers.

Very fast communications could reduce the need for big on-chip caches, freeing up real estate for other processing functions, Drost says.

The result: The technology could enable processor chips to communicate 60 times faster and with 30 times less energy than is possible using conventional circuit boards.

Proximity communication relies on chips that are placed very close together -- a tough configuration to manufacture in high volumes at a low cost.

To compensate for misalignment between chips, Sun researcher Robert Bosnyak and Chief Technology Officer Greg Papadopoulos developed a technique called electronic alignment that allows the transmitters and receivers on each chip to shift positions to compensate for a few hundred microns of misalignment.

"That was absolutely critical," says Drost. "Before that, [proximity communication] seemed really interesting but impractical."

Drost says he expects proximity communication to greatly improve the speed of applications requiring high-performance computing, such as weather modeling, car-crash simulations or drug simulations.

In 2003, the U.S. government awarded Sun Labs a US$50 million Defense Advanced Research Projects Agency contract to help design next-generation supercomputers. That project will be completed in 2010, but proximity communication technology will show up in other applications before then, Drost says. When that happens, he adds, "it's a whole different ballgame."

Collett is a Computerworld contributing writer. Contact her at stcollett@aol.com.

Click here to login and leave a comment.  

0 Comments

Add Comment

Text Only 2000 character limit

Page 1 of 1

White Papers & Special Reports

thomson reuters image
Biomarkers: An Indispensible Addition to the Drug Development Toolkit
Examining the Potential of Biomarkers
Sponsored by Thomson Reuters

Biomarkers are becoming an essential part of clinical development. In this white paper, Thomson Reuters provides insight from experts in industry and academia, and explores the role of biomarkers as evaluative tools in improving clinical research and the challenges this presents.

Discover the potential of biomarkers to:

  • Improve decision making
  • Accelerate drug development
  • Reduce development costs


BlueArc_Scientific Data
Scientific Data Lifecycle Management: Preparing for Storage in an Uncertain Future
Sponsored by BlueArc

Managing vast and overwhelming streams of gene sequencing data today requires ultra-high performance systems and processes. With continued rapid advancement and improvements in gene sequencing, expect tomorrow’s instruments to output quantities of genomic information that will dwarf current levels. Help your organization maintain data control and prepare for the future of sequencing through this informative paper that discusses:

  • The information technology challenges of gene sequencing
  • “Intelligent” methods for data management and customization
  • System survival tips... Deciding what data to keep or delete
  • New tools to keep scientists ahead of impending data torrents


SAS Managed image
Managed Innovation, Assured Compliance
Developing, executing and managing the transformation, analysis and submission of clinical research data with SAS® Drug Development
Sponsored by SAS
Get better products to market faster. Download this white paper to discover the top ten challenges facing life science executives and how to overcome them. See how SAS Drug Development transforms clinical data into true innovation.


Life Science Webcasts & Podcasts

Presented by Trade Commission of Spain

Spain Biotech: An Engine for Economic Change 

TCS podcastDiscover how Spain is focusing on biotechnology to be an engine for economic change through gradual internationalization, development and technology transfer.

Regional governments are actively investing in public and private biology research and promoting the creation of knowledge-based companies. Spain’s human capital combined with aggressive investment in biotech research and infrastructure has led to the creation of bio-clusters.

Today, there are nearly 700 Spanish companies engaged in biotechnology, with almost 50 percent growth in funding devoted to research. In fact, spending on internal R & D in biotechnology has grown 46 percent and is close to 300 million Euros.

Access the podcast 

 



More Podcasts

Job Openings

saic_logo

MANAGER, SCIENTIFIC COMPUTING & PROGRAMMING
(Bioinformatics Manager)
SAIC-Frederick, Inc has an exciting opportunity for a Manager, Scientific Computing & Programming - Core Genoytyping Facility in Gaithersburg, Maryland.  In this role, you will lead the Bioinformatics & Analysis Group.
Master’s or equivalent required.  PhD preferred. Six years experience in development of scientific programs in high-performance computing environment including five years supporting scientific research in computational chemistry, biology, or genetics, & two years supervisory experience.  View complete job posting & apply: www.saic-frederick.com. Position #146945.

For reprints and/or copyright permission, please contact The YGS Group, 1808 Colonial Village Lane, Lancaster, PA;

(717) 399-1900 ext. 125, or via email to Ashley.Zander@theYGSgroup.com.