Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 67
D
Findings and Recommendations from The Future of
Computing Performance: Game Over or Next Level?
T he following findings and recommendations are
repeated from the National Research Council's
report, The Future of Computing Performance:
Game Over or Next Level?1
·
systems not only by parallel-systems experts but also
by typical programmers.
Focus long-term efforts on rethinking of the canonical
computing "stack"--applications, programming
language, compiler, runtime, virtual machine,
Findings: operating system, hypervisor, and architecture--in
· The information technology sector itself and most light of parallelism and resource-management
other sectors of society--for example, manufacturing, challenges.
financial and other services, science, engineering, · Invest in research on and development of parallel
education, defense and other government services, and architectures driven by applications, including
entertainment--have grown dependent on continued enhancements of chip multiprocessor systems and
growth in computing performance. conventional data-parallel architectures, cost-effective
· After many decades of dramatic exponential growth, designs for application-specific architectures, and
single processor performance is increasing at a much support for radically different approaches.
lower rate, and this situation is not expected to · Invest in research and development to make computer
improve in the foreseeable future. systems more power-efficient at all levels of the
· The growth in the performance of computing system, including software, application-specific
systems--even if they are multiple-processor parallel approaches, and alternative devices. Such efforts
systems--will become limited by power consumption should address ways in which software and system
within a decade. architectures can improve power efficiency, such as by
· There is no known alternative to parallel systems for exploiting locality and the use of domain-specific
sustaining growth in computing performance; execution units. R&D should also be aimed at making
however, no compelling programming paradigms for logic gates more power-efficient. Such efforts should
general parallel systems have yet emerged. address alternative physical devices beyond
incremental improvements in today's CMOS circuits.
Recommendations: · To promote cooperation and innovation by sharing,
· Invest in research in and development of algorithms encourage development of open interface standards for
that can exploit parallel processing. parallel programming rather than proliferating
· Invest in research in and development of programming proprietary programming environments.
methods that will enable efficient use of parallel · Invest in the development of tools and methods to
transform legacy applications to parallel systems.
· Incorporate in computer science education an
1
increased emphasis on parallelism, and use a variety of
NRC, 2011, The Future of Computing Performance: Game
methods and approaches to better prepare students for
Over or Next Level?, Washington, D.C.: The National Academies
Press (available online at http://www.nap.edu/catalog.php?record_ the types of computing resources that they will
id=12980). encounter in their careers.
67
OCR for page 68
