Ignore:
Timestamp:
Feb 10, 2015, 5:55:14 PM (4 years ago)
Author:
cameron
Message:

Intro and Background, remove section 3

File:
1 edited

Legend:

Unmodified
Added
Removed
  • docs/Working/icGrep/introduction.tex

    r4489 r4490  
    11\section{Introduction}
    22
    3 Unix {\tt grep} is a tool widely used
    4 to search for lines in text files matching a given regular expression
    5 pattern.   Historical comments ...
    63
    7 Unicode regular expression matching adds performance challenges...
     4Although well-established technical standards exist for
     5Unicode regular expressions \cite{davis2012unicode}, most of today's
     6regular expression processing toolsets fail to support the full set of processing
     7features even at the most basic level \cite{stewart2013unicode}. 
     8One of the fundamental issues is performance and so it makes good sense
     9to consider the ways in which parallel processing approaches can help
     10address the gap.
    811
    912Efforts to improve the performance of regular expression matching through
     
    3336require thousands of DFA states for named Unicode properties.
    3437Building on the Parabix framework, Cameron et al.~\cite{cameron2014bitwise} introduce
    35 regular expression matching using the bitwise
    36 data parallel approach together with the MatchStar primitive
    37 for efficient implementation of
    38 Kleene-* character-class repetitions.
     38regular expression matching using a new bitwise
     39data parallel approach.
    3940
    4041In this paper, we report on the use of the implementation of a full
     
    4445The result is \icGrep{},
    4546a high-performance, full-featured open-source grep implementation
    46 with systematic support for Unicode regular expressions addressing the
    47 requirements of Unicode Technical Standard \#18 \cite{davis2012unicode}.  As an alternative
     47with systematic support for Unicode regular expressions.  As an alternative
    4848to classical grep implementations, \icGrep{} offers dramatic performance
    49 acceleration in ASCII-based and Unicode matching performance alike.
     49acceleration in Unicode regular expression matching.
    5050
    5151The remainder of this paper is organized as follows.   Section \ref{sec:background}
     
    5454the Parabix framework and regular expression matching techniques
    5555using bitwise data parallelism.   
    56 Section \ref{sec:seqpar} expands on previous work on bitwise data
    57 parallelism by more fully characterizing the paradigm and documenting
    58 important techniques.
     56%Section \ref{sec:seqpar} expands on previous work on bitwise data
     57%parallelism by more fully characterizing the paradigm and documenting
     58%important techniques.
    5959Section \ref{sec:Unicode} addresses
    6060the issues and performance challenges associated with meeting Unicode
Note: See TracChangeset for help on using the changeset viewer.