Skip to main content

From Maps to Models Augmenting the Nation's Geospatial Intelligence Capabilities (2016) / Chapter Skim
Currently Skimming:

Appendix B Computation
Pages 113-116

The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.


From page 113...
... Owing to specialized hardware, such as memory technologies capable of continuously feeding the vector registers, these machines were expensive, and so computations per dollar remained on the same power-law curve. This power-law behavior is often referred to as Moore's law, based on Gordon Moore's observation that transistor density on a chip doubled roughly every 18 months (Chien and Karamcheti, 2013; Kogge et al., 2008)
Read the entire page →
From page 114...
... The increase in computing speed made possible by the miniaturization of complementary metal oxide semiconductor technology is coming to an end, and speed increases now come from executing many instructions in parallel. When two concurrent computations use the same memory address, there is danger of a race condition, in which the results depend on the order of completion of operations.
Read the entire page →
From page 115...
... A standard programming model for thread-based execution is OpenMP, which places directives or pragmas inside code indicating that a particular code section allows thread-safe parallel execution. Parallel execution streams in distributed memory are called processes, and the standard programming model is the message-passing interface (MPI)
Read the entire page →
From page 116...
... 116 FROM MAPS TO MODELS MapReduce, such as the Hadoop distributed file system, are an emerging alternative to the standard POSIX-based file systems of hierarchical directories and files common on commodity platforms.
Read the entire page →

Key Terms

  • parallel computing
  • random access
  • vector registers
  • distributed memory
  • vector processing

  • access memory
  • vector length
  • parallel execution
  • vector machines
  • dennard scaling

  • address space
  • mass market
  • programming model
  • global address
  • symmetric multiprocessor

  • processing unit
  • standard programming
  • data streams
  • speed increases
  • conventional silicon

  • graphical processing
  • contain data
  • hierarchy spanning
  • mobile market
  • storage called

  • allow distributed
  • using commodity
  • processors contain
  • nonvolatile random
  • recent processor


    • This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
    More information on Chapter Skim is available.