# source:docs/ICA3PP2015/cameron2015bitwise.tex

Last change on this file was 4923, checked in by cameron, 3 years ago

Archiving ICA3PP2015 author version

File size: 2.7 KB
Line
2\usepackage{tikz}
4\usepackage{pgfplots}
5\usepackage{pgfplotstable}
6\usepackage{booktabs,siunitx,ragged2e}
7\usepackage{subfig}
8\usepackage{amsmath}
9\newcommand{\BigO}[1]{\ensuremath{\operatorname{O}\bigl(#1\bigr)}}
10\usepackage[official]{eurosym}
11\usepackage[utf8]{inputenc}
12\usepackage{listings}
13\lstset{breaklines=true,basicstyle=\small\ttfamily}
14\usepackage{bbding}
15
16\newcommand{\icGrep}[1]{icGrep}
17
18\def\RegularExpression{RegEx}
19\def\Pablo{Parabix}
20\def\CodeUnit{Code Unit}
21\def\REParser{\RegularExpression{} Parser}
22\def\CodeUnitCompiler{\CodeUnit{} Compiler}
23\def\RegularExpressionCompiler{\RegularExpression{} Compiler}
24\def\PabloCompiler{\Pablo{} Compiler}
25
26\pagestyle{empty}
27
28\begin{document}
29
30\title{Bitwise Data Parallelism with LLVM: The ICgrep Case Study}
31\author{Robert D. Cameron\Envelope
32\and Nigel Medforth
33\and Dan Lin
34\and Dale Denis
35\and William N. Sumner
36}
37\institute{School of Computing Science, Simon Fraser University, Surrey, B.C., Canada\\
38{\tt \{cameron,nmedfort,lindanl,daled,wsumner\}@sfu.ca}}
39%\authorrunning{Cameron, Medforth, Lin, Denis and Sumner}
40\maketitle
41
42
43\begin{abstract}
44Bitwise data parallelism using short vector (SIMD) instructions has recently been shown to have considerable promise
45as the basis for a new, fundamentally parallel, style of regular expression
46processing.
47This paper examines the application of this
48approach to the development a full-featured Unicode-capable open-source grep
49implementation.  Constructed using a layered architecture
50combining Parabix and LLVM compiler technologies, icGrep is the
51first instance of a potentially large class of text processing
52applications that achieve high performance text processing through the combination
53of dynamic compilation and bitwise data parallelism.
54In performance comparisons with several contemporary alternatives,
5510$\times$ or better speedups are often observed.
56\vskip 10pt
57This is the authors' version of the
58paper published in Algorithms and Architectures for Parallel Processing,
59Wang, Guojun and Zomaya, Albert and Perez, Gregorio Martinez and Li, Kenli (eds),
60{\em Lecture Notes in Computer Science} {\b 9529}, Nov. 2015, pp 373-387,
61http://dx.doi.org/10.1007/978-3-319-27122-4\_26.
62The final publication is available at link.springer.com.
63
64
65\keywords{
66Bitwise Data Parallelism;
67Dynamic Compilation;
68ICgrep;
69MatchStar;
70Parabix Transform;
71Regular Expression;
72SIMD
73}
74
75\end{abstract}
76
77\input{introduction.tex}
78
79\input{background.tex}
80
81\input{unicode-re.tex}
82
83\input{architecture.tex}
84
85\input{evaluation.tex}
86
87\input{conclusion.tex}
88
89\bibliographystyle{splncs}
90\bibliography{bitgrep}
91\end{document}
Note: See TracBrowser for help on using the repository browser.