December 17, 2010
WASHINGTON, Dec. 16 -- The rapid advances in information technology that drive many sectors of the US economy could stall unless the nation aggressively pursues fundamental research and development of parallel computing -- hardware and software that enable multiple computing activities to process simultaneously, says a new report by the National Research Council. Better options for managing power consumption in computers will also be essential for continued improvements in IT performance.
For many decades, advances in single-processor, sequential computer microprocessors have enabled computing performance to increase dramatically -- on the order of 10,000 times in the last 20 years alone. However, power management and other technological limitations have made it impractical to continue improving computer performance in this way much longer. Parallel computing, therefore, is the only known alternative for improving computer performance without significantly increasing costs and energy usage, the report says.
"The societal and economic impact of computer technology is undeniable, increasing productivity and efficiency and fostering innovation in medicine, defense, entertainment, and communications," said Samuel H. Fuller, chief technology officer and vice president of research and development for Analog Devices Inc., Norwood, Mass., and chair of the committee that wrote the report. "To ensure that computing systems continue to double in performance every few years, we need to make significant changes in computer software and hardware. Investing in research and development of parallel computing offers a clear path forward."
Despite some mainstream successes in parallel computing -- such as the MapReduce programming framework used by Google to process large data sets using thousands of computers -- most parallel computing in use now is limited to comparatively narrow scientific and engineering applications. To enable parallel computing for broader use, new algorithms, programming models, operating systems, and computer architectures will be required, the report says, and research and development in these areas should be pursued.
In particular, advances are necessary to develop new parallel programming methods and supporting computing systems. Although computing hardware such as semiconductor chips that contain eight or more microprocessors have already been developed, software that can keep that many or more processors busy in parallel is not available for most computing applications.
Research and development should also focus on making computer systems more energy efficient, the report says. Power constraints now affect systems ranging from handheld devices to the largest computing datacenters. Most computer chips are designed with silicon-based complementary metal oxide semiconductor (CMOS) technology. While the number of devices per CMOS chip continues to double every few years, the technology has essentially reached its limits with regard to power efficiency. Even as new parallel computing models and solutions are found, most future performance will ultimately be limited by energy constraints, the report notes.
It cautions that while parallel computing is the best alternative for improving future performance, there is no guarantee that it will bring rapid advances like those experienced in recent decades, and a number of uncertainties still need to be addressed. Therefore, research and development should also explore fundamentally different alternatives to today's CMOS technology.
The report also recommends developing open interface standards for parallel programming to promote cooperation and innovation in the industry, designing tools and methods for transferring today's sequential computing to parallel applications, and emphasizing parallel computing as part of computer science education.
The study was sponsored by the National Science Foundation. The National Academy of Sciences, National Academy of Engineering, Institute of Medicine, and National Research Council make up the National Academies. They are private, nonprofit institutions that provide science, technology, and health policy advice under a congressional charter. The Research Council is the principal operating agency of the National Academy of Sciences and the National Academy of Engineering. For more information, visit http://national-academies.org.
Molly Galvin, Media Relations Officer
Christopher White, Media Relations Assistant
Office of News and Public Information
202-334-2138; e-mail email@example.com
Pre-publication copies of The Future of Computing Performance: Game Over or Next Level? are available from the National Academies Press; tel. 202-334-3313 or 1-800-624-6242 or on the Internet at http://www.nap.edu. Reporters may obtain a copy from the Office of News and Public Information (contacts listed above).
Source: National Academy of Sciences
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?