Citation
Edward A. Lee. "Position Statement for Panel on Grand Challenges in Embedded Software". EMSOFT 07, Salzburg, Austria, October, 2007.

Abstract
Abstractions currently used in computing hide timing properties of software. As a consequence, computer scientists have developed techniques that deliver improved average-case performance and/or design convenience at the expense of timing predictability. For embedded software, which interacts closely with physical processes, timing is usually an essential property. Lack of timing in the core abstractions results is brittle and non-portable designs. Moreover, as embedded software becomes more networked, the prevailing empirical test-based approach to achieving real-time computing becomes inadequate. I believe it is necessary to reintroduce timing predictability as a first-class property of embedded processor architectures. Architectures currently strive for superior average case performance that regrettably ignores predictability and repeatability of timing properties. "Correct" execution of a C program has nothing to do with how long it takes to perform any particular action. C says nothing about timing, so timing is not considered part of correctness. Architectures have developed deep pipelines with speculative execution and dynamic dispatch. Memory architectures have developed multi-level caches and TLBs. The performance criterion is simple: faster (on average) is better.

Electronic downloads

• grandChallenges_Lee.pdf · · 71 kbytes

• HTML

  Edward A. Lee. <a
  href="http://chess.eecs.berkeley.edu/pubs/348.html"
  >Position Statement for Panel on Grand Challenges in
  Embedded Software</a>, EMSOFT 07, Salzburg, Austria,
  October, 2007.

• Plain text

  Edward A. Lee. "Position Statement for Panel on Grand
  Challenges in Embedded Software". EMSOFT 07, Salzburg,
  Austria, October, 2007.

• BibTeX

  @inproceedings{Lee07_PositionStatementForPanelOnGrandChallengesInEmbedded,
      author = {Edward A. Lee},
      title = {Position Statement for Panel on Grand Challenges
                in Embedded Software},
      booktitle = {EMSOFT 07, Salzburg, Austria},
      month = {October},
      year = {2007},
      abstract = {Abstractions currently used in computing hide
                timing properties of software. As a consequence,
                computer scientists have developed techniques that
                deliver improved average-case performance and/or
                design convenience at the expense of timing
                predictability. For embedded software, which
                interacts closely with physical processes, timing
                is usually an essential property. Lack of timing
                in the core abstractions results is brittle and
                non-portable designs. Moreover, as embedded
                software becomes more networked, the prevailing
                empirical test-based approach to achieving
                real-time computing becomes inadequate. I believe
                it is necessary to reintroduce timing
                predictability as a first-class property of
                embedded processor architectures. Architectures
                currently strive for superior average case
                performance that regrettably ignores
                predictability and repeatability of timing
                properties. "Correct" execution of a C program has
                nothing to do with how long it takes to perform
                any particular action. C says nothing about
                timing, so timing is not considered part of
                correctness. Architectures have developed deep
                pipelines with speculative execution and dynamic
                dispatch. Memory architectures have developed
                multi-level caches and TLBs. The performance
                criterion is simple: faster (on average) is
                better. },
      URL = {http://chess.eecs.berkeley.edu/pubs/348.html}
  }
  

Posted by Christopher Brooks on 1 Oct 2007.
For additional information, see the Publications FAQ or contact webmaster at chess eecs berkeley edu.