Changes between Version 17 and Version 18 of WikiStart

May 14, 2012, 7:46:30 AM (7 years ago)



  • WikiStart

    v17 v18  
    1 = Welcome to the Development Home of Parabix - World's Fastest XML Software =
    3 Parabix (Parallel bit streams for XML - TM) demonstrates a fundamentally new way to
    4 perform high-speed parsing of XML documents.   Byte-oriented character stream data
     2Parabix technology is a high-performance programming framework for streaming
     3text processing applications, leveraging both SIMD and multicore
     4parallel processing features. 
     6== Parallel Bit Streams ==
     8The Parabix framework is based on the concept of parallel bit streams,
     9a fundamentally new transform representation of text.   Byte-oriented character stream data
    510is first transformed into eight parallel bit streams, each bit stream comprising
    611one bit per character code unit.  Code units may be ASCII characters or
    1116to process 128 code unit positions at a time.
    13 Applications of parallel bit stream technology to various text/Unicode/XML
    14 processing applications is the ongoing research of Prof. Rob Cameron of the
    15 School of Computing Science at Simon Fraser University.  See also the
    16 [ u8u16] project for application of parallel bit
    17 stream technology to high-speed UTF-8 to UTF-16 transcoding.
     18== The Pablo Language and Compilers ==
     20The PabloLanguage allows parallel bit stream programs to be conveniently
     21written using primitives that manipulate arbitrary-length bitstreams.
     24== IDISA Run-Time Libraries ==
     26The [wiki:IDISAproject IDISA] project defines an abstraction for
     27portable SIMD programming featuring support for operations at all
     28power-of-2 field widths as well as transitions between those field-widths
     29(inductive doubling architecture).
     32== Commercial Applications ==
    1934Parabix software as provided here as open source software under Open Software
    26 == Parabix 2 ==
    28 Our research program is presently moving on to the development of
    29 Parabix 2 - a second version of Parabix with four technical advances
    30 over our first version.
    32  1. Parallel Parsing with Bitstream Addition
    34  This is a new technique which is demonstrated in our
    35  python [source:/proto prototype].
    37  2. Compiler Technology
    39  We are developing compiler technology to automatically generate
    40  much of the low-level SIMD code for parallel bit stream
    41  processing.  Although the first version of Parabix used
    42  character class compilation to produce bitlex.c, Parabix 2
    43  will employ compilation techniques much more broadly.
    45  3. Array Set Models
    47  Array set models are a technique for representing XML
    48  infomation items using sets of arrays.  The primary goal is
    49  to enable the use of high-performance techniques in Java
    50  through bulk array transport across the JNI boundary.
    52  4. Length-Sorted Symbol Tables
    54  By using bit-scan instructions to cheaply determine the length
    55  of names without touching each byte, name occurrences can
    56  first be presorted according to length.  Separate loops
    57  or routines are then used for processing the name occurrences
    58  of each length; these routines can simply load and compare
    59  all the bytes of names without byte-at-a-time loops.
    61 Cameron, Rob, Ken Herdy and Ehsan Amiri.
    62 [ Parallel Bit Stream Technology as a Foundation for XML Parsing Performance]. Presented at International Symposium on Processing XML Efficiently: Overcoming Limits on Space, Time, or Bandwidth, Montréal, Canada, August 10, 2009. In Proceedings of the International Symposium on Processing XML Efficiently: Overcoming Limits on Space, Time, or Bandwidth. Balisage Series on Markup Technologies, vol. 4 (2009). doi:10.4242/BalisageVol4.Cameron01.
    64 Robert D. Cameron, Kenneth S. Herdy and Dan Lin,
    65 [ High Performance XML Parsing Using Parallel Bit Stream Technology],
    66 Proceedings of CASCON 2008, Toronto, Ontario, October 27-30, 2008.
    68 Kenneth S. Herdy, David S. Burggraf, and Robert D. Cameron,
    69 [ High Performance GML to SVG Transformation for the Visual Presentation of Geographic Data in Web-Based Mapping Systems],
    70 Proceedings of SVG Open, Nuremberg, Germany, August 26-28, 2008.
    72 January 1, 2010: XmlStringValueExtraction
    74 [wiki:IDISAproject IDISA Project]