source: docs/HPCA2012/10-related.tex @ 1394

Last change on this file since 1394 was 1394, checked in by ashriram, 8 years ago

Related work fixed

File size: 1.8 KB
Line 
1\section{Related Work}
2\label{section:related}
3
4There has been work in the past which has sought to address the
5overheads of text processing in specific applications (e.g., XML
6parsers) and have adopted specialized hardware and software solutions
7for each application.
8% Event-based SAX (Simple API for XML) parsers avoid the tree
9% construction costs of the more flexible DOM (Document Object Model)
10% parsers \cite{Perkins05}.
11Nicola and John specifically identified the traditional method of XML
12parsing as a threat to database performance and outlined a number of
13potential directions for improving performance \cite{NicolaJohn03}.
14The commercial importance of XML parsing has spurred the development
15of numerous multi-threaded and hardware-based approaches:
16Multithreaded XML techniques include preparsing the XML file to locate
17key partitioning points \cite{ZhangPanChiu09} and speculative p-DFAs
18\cite{ZhangPanChiu09}. Hardware methods include custom XML chips
19\cite{Leventhal2009} and FPGA-based implementations
20\cite{DaiNiZhu2010}.  Recently Cameron et
21al.~\cite{CameronHerdyLin2008, cameron-EuroPar2011} accelerated XML
22parsing using SSE instructions. Finally, other have explored the
23design of custom hardware for bit parallel operations in network
24processors~\cite{tan-sherwood-isca-2005}.
25
26
27In this paper, we have introduce parallel bit streams as a general
28abstraction to parallelize and improve the performance general text
29processing. We have developed a compiler tool chain and the runtime to
30enable bit streams to exploit SIMD extensions found on commodity
31processors.  We are also the first to perform a detailed analysis of
32SIMD instruction extensions across three generations of intel
33processors including the new 256 bit AVX extensions. Finally, we have
34shown the benefits of using multithreading in conjunction with data
35parallel phases of the application.
Note: See TracBrowser for help on using the repository browser.