Changeset 3055 for docs/Balisage13


Ignore:
Timestamp:
Apr 19, 2013, 4:27:25 PM (6 years ago)
Author:
cameron
Message:

Linking

Location:
docs/Balisage13/Bal2013came0601
Files:
2 edited

Legend:

Unmodified
Added
Removed
  • docs/Balisage13/Bal2013came0601/Bal2013came0601.html

    r3053 r3055  
    274274</div>
    275275<div id="mast"><div class="content">
    276 <h2 class="article-title" id="idp66752"></h2>
     276<h2 class="article-title" id="idp66368"></h2>
    277277<div class="author">
    278278<h3 class="author">Nigel Medforth</h3>
     
    324324</div>
    325325<div class="mast-box">
    326 <p class="title"><a href="javascript:toggle('idp67872')" class="quiet"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp67872"></a> <span onclick="javascript:toggle('idp67872');return true">Abstract</span></p>
    327 <div class="folder" id="folder-idp67872" style="display:none"><p id="idp68176">Prior research on the acceleration of XML processing using SIMD and multi-core
     326<p class="title"><a href="javascript:toggle('idp67488')" class="quiet"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp67488"></a> <span onclick="javascript:toggle('idp67488');return true">Abstract</span></p>
     327<div class="folder" id="folder-idp67488" style="display:none"><p id="idp67792">Prior research on the acceleration of XML processing using SIMD and multi-core
    328328            parallelism has lead to a number of interesting research prototypes. This work
    329329            investigates the extent to which the techniques underlying these prototypes could result
     
    339339<p><b>Table of Contents</b></p>
    340340<dl>
    341 <dt><span class="section"><a href="#idp275888" class="toc">Introduction</a></span></dt>
    342 <dt><span class="section"><a href="#idp277680" class="toc">Background</a></span></dt>
     341<dt><span class="section"><a href="#idp276064" class="toc">Introduction</a></span></dt>
     342<dt><span class="section"><a href="#idp277856" class="toc">Background</a></span></dt>
    343343<dd><dl>
    344 <dt><span class="section"><a href="#idp278320" class="toc">Xerces C++ Structure</a></span></dt>
    345 <dt><span class="section"><a href="#idp321360" class="toc">The Parabix Framework</a></span></dt>
    346 <dt><span class="section"><a href="#idp411088" class="toc">Sequential vs. Parallel Paradigm</a></span></dt>
     344<dt><span class="section"><a href="#idp278496" class="toc">Xerces C++ Structure</a></span></dt>
     345<dt><span class="section"><a href="#idp333696" class="toc">The Parabix Framework</a></span></dt>
     346<dt><span class="section"><a href="#idp426176" class="toc">Sequential vs. Parallel Paradigm</a></span></dt>
    347347</dl></dd>
    348 <dt><span class="section"><a href="#idp415472" class="toc">Architecture</a></span></dt>
     348<dt><span class="section"><a href="#idp430592" class="toc">Architecture</a></span></dt>
    349349<dd><dl>
    350 <dt><span class="section"><a href="#idp416112" class="toc">Overview</a></span></dt>
    351 <dt><span class="section"><a href="#idp444144" class="toc">Character Set Adapters</a></span></dt>
    352 <dt><span class="section"><a href="#idp451856" class="toc">Combined Parallel Filtering</a></span></dt>
    353 <dt><span class="section"><a href="#idp467808" class="toc">Content Stream</a></span></dt>
    354 <dt><span class="section"><a href="#idp478528" class="toc">Namespace Handling</a></span></dt>
    355 <dt><span class="section"><a href="#idp521984" class="toc">Error Handling</a></span></dt>
     350<dt><span class="section"><a href="#idp431264" class="toc">Overview</a></span></dt>
     351<dt><span class="section"><a href="#idp460576" class="toc">Character Set Adapters</a></span></dt>
     352<dt><span class="section"><a href="#idp468560" class="toc">Combined Parallel Filtering</a></span></dt>
     353<dt><span class="section"><a href="#idp485632" class="toc">Content Stream</a></span></dt>
     354<dt><span class="section"><a href="#idp496352" class="toc">Namespace Handling</a></span></dt>
     355<dt><span class="section"><a href="#idp539680" class="toc">Error Handling</a></span></dt>
    356356</dl></dd>
    357 <dt><span class="section"><a href="#idp532208" class="toc">Multithreading with Pipeline Parallelism</a></span></dt>
    358 <dt><span class="section"><a href="#idp554320" class="toc">Performance</a></span></dt>
     357<dt><span class="section"><a href="#idp549968" class="toc">Multithreading with Pipeline Parallelism</a></span></dt>
     358<dt><span class="section"><a href="#idp572016" class="toc">Performance</a></span></dt>
    359359<dd><dl>
    360 <dt><span class="section"><a href="#idp557040" class="toc">Xerces C++ SAXCount</a></span></dt>
    361 <dt><span class="section"><a href="#idp581280" class="toc">GML2SVG</a></span></dt>
     360<dt><span class="section"><a href="#idp574736" class="toc">Xerces C++ SAXCount</a></span></dt>
     361<dt><span class="section"><a href="#idp598880" class="toc">GML2SVG</a></span></dt>
    362362</dl></dd>
    363 <dt><span class="section"><a href="#idp598304" class="toc">Conclusion and Future Work</a></span></dt>
     363<dt><span class="section"><a href="#idp624032" class="toc">Conclusion and Future Work</a></span></dt>
    364364</dl>
    365365</div>
    366366<div class="mast-box">
    367 <p class="title"><a href="javascript:toggle('idp69600')" class="linkbox"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp69600"></a> <span onclick="javascript:toggle('idp69600');return true">Nigel Medforth</span></p>
    368 <div class="folder" id="folder-idp69600" style="display:none">
     367<p class="title"><a href="javascript:toggle('idp69216')" class="linkbox"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp69216"></a> <span onclick="javascript:toggle('idp69216');return true">Nigel Medforth</span></p>
     368<div class="folder" id="folder-idp69216" style="display:none">
    369369<h5 class="author-email"><code class="email">&lt;<a class="email" href="mailto:nmedfort@sfu.ca">nmedfort@sfu.ca</a>&gt;</code></h5>
    370370<div class="affiliation">
     
    377377</div>
    378378<div class="personblurb">
    379 <p id="idp51296">Nigel Medforth is a M.Sc. student at Simon Fraser University and the lead
     379<p id="idp50976">Nigel Medforth is a M.Sc. student at Simon Fraser University and the lead
    380380               developer of icXML. He earned a Bachelor of Technology in Information Technology at
    381381               Kwantlen Polytechnic University in 2009 and was awarded the Dean’s Medal for
    382382               Outstanding Achievement.</p>
    383 <p id="idp52304">Nigel is currently researching ways to leverage both the Parabix framework and
     383<p id="idp51984">Nigel is currently researching ways to leverage both the Parabix framework and
    384384               stream-processing models to further accelerate XML parsing within icXML.</p>
    385385</div>
     
    387387</div>
    388388<div class="mast-box">
    389 <p class="title"><a href="javascript:toggle('idp55968')" class="linkbox"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp55968"></a> <span onclick="javascript:toggle('idp55968');return true">Dan Lin</span></p>
    390 <div class="folder" id="folder-idp55968" style="display:none">
     389<p class="title"><a href="javascript:toggle('idp55648')" class="linkbox"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp55648"></a> <span onclick="javascript:toggle('idp55648');return true">Dan Lin</span></p>
     390<div class="folder" id="folder-idp55648" style="display:none">
    391391<h5 class="author-email"><code class="email">&lt;<a class="email" href="mailto:lindanl@sfu.ca">lindanl@sfu.ca</a>&gt;</code></h5>
    392392<div class="affiliation">
     
    394394<p class="orgname">Simon Fraser University </p>
    395395</div>
    396 <div class="personblurb"><p id="idp57680">Dan Lin is a Ph.D student at Simon Fraser University. She earned a Master of Science
     396<div class="personblurb"><p id="idp57360">Dan Lin is a Ph.D student at Simon Fraser University. She earned a Master of Science
    397397             in Computing Science at Simon Fraser University in 2010. Her research focus on on high
    398398             performance algorithms that exploit parallelization strategies on various multicore platforms.
     
    401401</div>
    402402<div class="mast-box">
    403 <p class="title"><a href="javascript:toggle('idp60240')" class="linkbox"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp60240"></a> <span onclick="javascript:toggle('idp60240');return true">Kenneth Herdy</span></p>
    404 <div class="folder" id="folder-idp60240" style="display:none">
     403<p class="title"><a href="javascript:toggle('idp59920')" class="linkbox"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp59920"></a> <span onclick="javascript:toggle('idp59920');return true">Kenneth Herdy</span></p>
     404<div class="folder" id="folder-idp59920" style="display:none">
    405405<h5 class="author-email"><code class="email">&lt;<a class="email" href="mailto:ksherdy@sfu.ca">ksherdy@sfu.ca</a>&gt;</code></h5>
    406406<div class="affiliation">
     
    409409</div>
    410410<div class="personblurb">
    411 <p id="idp61968"> Ken Herdy completed an Advanced Diploma of Technology in Geographical Information
     411<p id="idp61648"> Ken Herdy completed an Advanced Diploma of Technology in Geographical Information
    412412               Systems at the British Columbia Institute of Technology in 2003 and earned a Bachelor
    413413               of Science in Computing Science with a Certificate in Spatial Information Systems at
    414414               Simon Fraser University in 2005. </p>
    415 <p id="idp262752"> Ken is currently pursuing PhD studies in Computing Science at Simon Fraser
     415<p id="idp262928"> Ken is currently pursuing PhD studies in Computing Science at Simon Fraser
    416416               University with industrial scholarship support from the Natural Sciences and
    417417               Engineering Research Council of Canada, the Mathematics of Information Technology and
     
    423423</div>
    424424<div class="mast-box">
    425 <p class="title"><a href="javascript:toggle('idp265488')" class="linkbox"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp265488"></a> <span onclick="javascript:toggle('idp265488');return true">Rob Cameron</span></p>
    426 <div class="folder" id="folder-idp265488" style="display:none">
     425<p class="title"><a href="javascript:toggle('idp265664')" class="linkbox"><img class="toc-icon" src="plus.png" alt="expand" id="icon-idp265664"></a> <span onclick="javascript:toggle('idp265664');return true">Rob Cameron</span></p>
     426<div class="folder" id="folder-idp265664" style="display:none">
    427427<h5 class="author-email"><code class="email">&lt;<a class="email" href="mailto:cameron@cs.sfu.ca">cameron@cs.sfu.ca</a>&gt;</code></h5>
    428428<div class="affiliation">
     
    434434<p class="orgname">International Characters, Inc.</p>
    435435</div>
    436 <div class="personblurb"><p id="idp267152">Dr. Rob Cameron is Professor of Computing Science and Associate Dean of Applied
     436<div class="personblurb"><p id="idp267328">Dr. Rob Cameron is Professor of Computing Science and Associate Dean of Applied
    437437               Sciences at Simon Fraser University. His research interests include programming
    438438               language and software system technology, with a specific focus on high performance
     
    450450<div id="main">
    451451<div class="article">
    452 <h2 class="article-title" id="idp66752"></h2>
    453 <div class="section" id="idp275888">
     452<h2 class="article-title" id="idp66368"></h2>
     453<div class="section" id="idp276064">
    454454<h2 class="title" style="clear: both">Introduction</h2>
    455 <p id="idp276528"></p>
    456 <p id="idp276784"></p>
    457 <p id="idp277040"></p>
    458 <p id="idp277296"></p>
    459 </div>
    460 <div class="section" id="idp277680">
     455<p id="idp276704"></p>
     456<p id="idp276960"></p>
     457<p id="idp277216"></p>
     458<p id="idp277472"></p>
     459</div>
     460<div class="section" id="idp277856">
    461461<h2 class="title" style="clear: both">Background</h2>
    462 <div class="section" id="idp278320">
     462<div class="section" id="idp278496">
    463463<h3 class="title" style="clear: both">Xerces C++ Structure</h3>
    464 <p id="idp278960"> The Xerces C++ parser is a widely-used standards-conformant
     464<p id="idp279136"> The Xerces C++ parser is a widely-used standards-conformant
    465465            XML parser produced as open-source software
    466466             by the Apache Software Foundation.
     
    473473            parsing using either pull parsing or SAX/SAX2 push-style parsing as well as a DOM
    474474            tree-based parsing interface. </p>
    475 <p id="idp280224">
     475<p id="idp280400">
    476476            Xerces,
    477477            like all traditional parsers, processes XML documents sequentially a byte-at-a-time from
     
    480480            state. This introduces implicit dependencies between the various tasks within the
    481481            application that make it difficult to optimize for performance. As a complex software
    482             system, no one feature dominates the overall parsing performance. Table I
     482              system, no one feature dominates the overall parsing performance. <a class="xref" href="#xerces-profile">Table I</a>
    483483            shows the execution time profile of the top ten functions in a
    484484            typical run. Even if it were possible, Amdahl's Law dictates that tackling any one of
     
    490490            expected that a comprehensive restructuring is required, involving all aspects of the
    491491            parser. </p>
    492 <div class="table-wrapper" id="idp283344">
     492<div class="table-wrapper" id="xerces-profile">
    493493<p class="title">Table I</p>
    494 <div class="caption"><p id="idm855552">Execution Time of Top 10 Xerces Functions</p></div>
    495 <table class="table">
     494<div class="caption"><p id="idm855808">Execution Time of Top 10 Xerces Functions</p></div>
     495<table class="table" xml:id="xerces-profile">
    496496<colgroup span="1">
    497497<col align="left" valign="top" span="1">
     
    547547</div>
    548548</div>
    549 <div class="section" id="idp321360">
     549<div class="section" id="idp333696">
    550550<h3 class="title" style="clear: both">The Parabix Framework</h3>
    551 <p id="idp322032"> The Parabix (parallel bit stream) framework is a transformative approach to XML
     551<p id="idp334368"> The Parabix (parallel bit stream) framework is a transformative approach to XML
    552552            parsing (and other forms of text processing.) The key idea is to exploit the
    553553            availability of wide SIMD registers (e.g., 128-bit) in commodity processors to represent
    554554            data from long blocks of input data by using one register bit per single input byte. To
    555555            facilitate this, the input data is first transposed into a set of basis bit streams.
    556               For example Table II shows  the ASCII bytes for the string "<code class="code">b7&lt;A</code>" with
    557                 the corresponding  8 basis bit streams, b<sub>0</sub> through  b<sub>7</sub> shown in Table III.
     556              For example, <a class="xref" href="#xml-bytes">Table II</a> shows  the ASCII bytes for the string "<code class="code">b7&lt;A</code>" with
     557                the corresponding  8 basis bit streams, b<sub>0</sub> through  b<sub>7</sub> shown in  <a class="xref" href="#xml-bits">Table III</a>.
    558558--&gt;
    559559           
     
    577577             multiple
    578578            classes can share the classification cost. </p>
    579 <div class="table-wrapper" id="idp334416">
     579<div class="table-wrapper" id="xml-bytes">
    580580<p class="title">Table II</p>
    581 <div class="caption"><p id="idp334928">XML Source Data</p></div>
    582 <table class="table">
     581<div class="caption"><p id="idp348928">XML Source Data</p></div>
     582<table class="table" xml:id="xml-bytes">
    583583<colgroup span="1">
    584584<col align="right" valign="top" span="1">
     
    606606</table>
    607607</div>
    608 <div class="table-wrapper" id="idp350320">
     608<div class="table-wrapper" id="xml-bits">
    609609<p class="title">Table III</p>
    610 <div class="caption"><p id="idp350832">8-bit ASCII Basis Bit Streams</p></div>
    611 <table class="table">
     610<div class="caption"><p id="idp365200">8-bit ASCII Basis Bit Streams</p></div>
     611<table class="table" xml:id="xml-bits">
    612612<colgroup span="1">
    613613<col align="centre" valign="top" span="1">
     
    675675</table>
    676676</div>
    677 <p id="idp390992"> Consider, for example, the XML source data stream shown in the first line of Table IV.
     677<p id="idp404848"> Consider, for example, the XML source data stream shown in the first line of <a class="xref" href="#derived">Table IV</a>.
    678678The remaining lines of this figure show
    679679            several parallel bit streams that are computed in Parabix-style parsing, with each bit
     
    687687            (using the technique of bitstream addition \cite{cameron-EuroPar2011}), namely streams
    688688            marking the element names, attribute names and attribute values of tags. </p>
    689 <div class="table-wrapper" id="idp393904">
     689<div class="table-wrapper" id="derived">
    690690<p class="title">Table IV</p>
    691 <div class="caption"><p id="idp394416">XML Source Data and Derived Parallel Bit Streams</p></div>
    692 <table class="table">
     691<div class="caption"><p id="idp408992">XML Source Data and Derived Parallel Bit Streams</p></div>
     692<table class="table" xml:id="derived">
    693693<colgroup span="1">
    694694<col align="centre" valign="top" span="1">
     
    739739</table>
    740740</div>
    741 <p id="idp406864"> Two intuitions may help explain how the Parabix approach can lead to improved XML
     741<p id="idp421952"> Two intuitions may help explain how the Parabix approach can lead to improved XML
    742742            parsing performance. The first is that the use of the full register width offers a
    743743            considerable information advantage over sequential byte-at-a-time parsing. That is,
     
    748748            individual decision-bits, an approach that computes many of them in parallel (e.g., 128
    749749            bytes at a time using 128-bit registers) should provide substantial benefit. </p>
    750 <p id="idp408112"> Previous studies have shown that the Parabix approach improves many aspects of XML
     750<p id="idp423200"> Previous studies have shown that the Parabix approach improves many aspects of XML
    751751            processing, including transcoding \cite{Cameron2008}, character classification and
    752752            validation, tag parsing and well-formedness checking. The first Parabix parser used
     
    757757            \cite{HPCA2012}. Although these research prototypes handled the full syntax of
    758758            schema-less XML documents, they lacked the functionality required by full XML parsers. </p>
    759 <p id="idp410240"> Commercial XML processors support transcoding of multiple character sets and can
     759<p id="idp425328"> Commercial XML processors support transcoding of multiple character sets and can
    760760            parse and validate against multiple document vocabularies. Additionally, they provide
    761761            API facilities beyond those found in research prototypes, including the widely used SAX,
    762762            SAX2 and DOM interfaces. </p>
    763763</div>
    764 <div class="section" id="idp411088">
     764<div class="section" id="idp426176">
    765765<h3 class="title" style="clear: both">Sequential vs. Parallel Paradigm</h3>
    766 <p id="idp411728"> Xerces—like all traditional XML parsers—processes XML documents
     766<p id="idp426816"> Xerces—like all traditional XML parsers—processes XML documents
    767767            sequentially. Each character is examined to distinguish between the XML-specific markup,
    768768            such as a left angle bracket <code class="code">&lt;</code>, and the content held within the
    769769            document. As the parser progresses through the document, it alternates between markup
    770770            scanning, validation and content processing modes. </p>
    771 <p id="idp413264"> In other words, Xerces belongs to an equivalent class applications termed FSM
     771<p id="idp428384"> In other words, Xerces belongs to an equivalent class applications termed FSM
    772772            applications\footnote{ Herein FSM applications are considered software systems whose
    773773            behaviour is defined by the inputs, current state and the events associated with
     
    775775            subsequent characters. Unfortunately, textual data tends to be unpredictable and any
    776776            character could induce a state transition. </p>
    777 <p id="idp414176"> Parabix-style XML parsers utilize a concept of layered processing. A block of source
     777<p id="idp429296"> Parabix-style XML parsers utilize a concept of layered processing. A block of source
    778778            text is transformed into a set of lexical bitstreams, which undergo a series of
    779779            operations that can be grouped into logical layers, e.g., transposition, character
     
    784784</div>
    785785</div>
    786 <div class="section" id="idp415472">
     786<div class="section" id="idp430592">
    787787<h2 class="title" style="clear: both">Architecture</h2>
    788 <div class="section" id="idp416112">
     788<div class="section" id="idp431264">
    789789<h3 class="title" style="clear: both">Overview</h3>
    790 <p id="idp417008"> icXML is more than an optimized version of Xerces. Many components were grouped,
     790<p id="idp432272"> icXML is more than an optimized version of Xerces. Many components were grouped,
    791791            restructured and rearchitected with pipeline parallelism in mind. In this section, we
    792792            highlight the core differences between the two systems. As shown in Figure
     
    814814<p class="title">Figure 1: Xerces Architecture</p>
    815815<div class="figure-contents">
    816 <div class="mediaobject" id="idp423824"><img alt="png image (xerces.png)" src="xerces.png" width="150cm"></div>
     816<div class="mediaobject" id="idp439440"><img alt="png image (xerces.png)" src="xerces.png" width="150cm"></div>
    817817<div class="caption"></div>
    818818</div>
    819819</div>
    820 <p id="idp426144"> In icXML functions are grouped into logical components. As shown in Figure
    821             \ref{fig:icxml-arch}, two major categories exist: (1) the Parabix Subsystem and (2) the
     820<p id="idp441808"> In icXML functions are grouped into logical components. As shown in Figure
     821             <a class="xref" href="#xerces-arch" title="Xerces Architecture">Figure 1</a>, two major categories exist: (1) the Parabix Subsystem and (2) the
    822822            Markup Processor. All tasks in (1) use the Parabix Framework \cite{HPCA2012}, which
    823823            represents data as a set of parallel bitstreams. The <span class="ital">Character Set
     
    837837            described in Section \ref{section:arch:errorhandling}. From here, two data-independent
    838838            branches exist: the Symbol Resolver and Content Preparation Unit. </p>
    839 <p id="idp430368"> A typical XML file contains few unique element and attribute names—but
     839<p id="idp446736"> A typical XML file contains few unique element and attribute names—but
    840840            each of them will occur frequently. icXML stores these as distinct data structures,
    841841            called symbols, each with their own global identifier (GID). Using the symbol marker
     
    843843               Resolver</span> scans through the raw data to produce a sequence of GIDs, called
    844844            the <span class="ital">symbol stream</span>. </p>
    845 <p id="idp432960"> The final components of the Parabix Subsystem are the <span class="ital">Content
     845<p id="idp449328"> The final components of the Parabix Subsystem are the <span class="ital">Content
    846846               Preparation Unit</span> and <span class="ital">Content Stream
    847847            Generator</span>. The former takes the (transposed) basis bitstreams and selectively
    848848            filters them, according to the information provided by the Parallel Markup Parser, and
    849849            the latter transforms the filtered streams into the tagged UTF-16 <span class="ital">content stream</span>, discussed in Section \ref{section:arch:contentstream}. </p>
    850 <p id="idp435872"> Combined, the symbol and content stream form icXML's compressed IR of the XML
     850<p id="idp452240"> Combined, the symbol and content stream form icXML's compressed IR of the XML
    851851            document. The <span class="ital">Markup Processor</span>~parses the IR to
    852852            validate and produce the sequential output for the end user. The <span class="ital">Final WF checker</span> performs inter-element well-formedness validation that
     
    860860<p class="title">Figure 2: icXML Architecture</p>
    861861<div class="figure-contents">
    862 <div class="mediaobject" id="idp441696"><img alt="png image (icxml.png)" src="icxml.png" width="500cm"></div>
     862<div class="mediaobject" id="idp458128"><img alt="png image (icxml.png)" src="icxml.png" width="500cm"></div>
    863863<div class="caption"></div>
    864864</div>
    865865</div>
    866866</div>
    867 <div class="section" id="idp444144">
     867<div class="section" id="idp460576">
    868868<h3 class="title" style="clear: both">Character Set Adapters</h3>
    869 <p id="idp444816"> In Xerces, all input is transcoded into UTF-16 to simplify the parsing costs of
     869<p id="idp461248"> In Xerces, all input is transcoded into UTF-16 to simplify the parsing costs of
    870870            Xerces itself and provide the end-consumer with a single encoding format. In the
    871871            important case of UTF-8 to UTF-16 transcoding, the transcoding costs can be significant,
     
    874874            other cases, transcoding may involve table look-up operations for each byte of input. In
    875875            any case, transcoding imposes at least a cost of buffer copying. </p>
    876 <p id="idp445872"> In icXML, however, the concept of Character Set Adapters (CSAs) is used to minimize
     876<p id="idp462304"> In icXML, however, the concept of Character Set Adapters (CSAs) is used to minimize
    877877            transcoding costs. Given a specified input encoding, a CSA is responsible for checking
    878878            that input code units represent valid characters, mapping the characters of the encoding
     
    880880            item streams), as well as supporting ultimate transcoding requirements. All of this work
    881881            is performed using the parallel bitstream representation of the source input. </p>
    882 <p id="idp446848"> An important observation is that many character sets are an extension to the legacy
     882<p id="idp463280"> An important observation is that many character sets are an extension to the legacy
    883883            7-bit ASCII character set. This includes the various ISO Latin character sets, UTF-8,
    884884            UTF-16 and many others. Furthermore, all significant characters for parsing XML are
    885885            confined to the ASCII repertoire. Thus, a single common set of lexical item calculations
    886886            serves to compute lexical item streams for all such ASCII-based character sets. </p>
    887 <p id="idp447728"> A second observation is that—regardless of which character set is
     887<p id="idp464160"> A second observation is that—regardless of which character set is
    888888            used—quite often all of the characters in a particular block of input will be
    889889            within the ASCII range. This is a very simple test to perform using the bitstream
     
    892892            be skipped. Transcoding to UTF-16 becomes trivial as the high eight bitstreams of the
    893893            UTF-16 form are each set to zero in this case. </p>
    894 <p id="idp449376"> A third observation is that repeated transcoding of the names of XML elements,
     894<p id="idp466080"> A third observation is that repeated transcoding of the names of XML elements,
    895895            attributes and so on can be avoided by using a look-up mechanism. That is, the first
    896896            occurrence of each symbol is stored in a look-up table mapping the input encoding to a
     
    899899            symbol look up is required to apply various XML validation rules, there is achieves the
    900900            effect of transcoding each occurrence without additional cost. </p>
    901 <p id="idp450432"> The cost of individual character transcoding is avoided whenever a block of input is
     901<p id="idp467136"> The cost of individual character transcoding is avoided whenever a block of input is
    902902            confined to the ASCII subset and for all but the first occurrence of any XML element or
    903903            attribute name. Furthermore, when transcoding is required, the parallel bitstream
     
    910910            using bit scan operations. </p>
    911911</div>
    912 <div class="section" id="idp451856">
     912<div class="section" id="idp468560">
    913913<h3 class="title" style="clear: both">Combined Parallel Filtering</h3>
    914 <p id="idp452544"> As just mentioned, UTF-8 to UTF-16 transcoding involves marking all but the last
     914<p id="idp469248"> As just mentioned, UTF-8 to UTF-16 transcoding involves marking all but the last
    915915            bytes of multi-byte UTF-8 sequences as positions for deletion. For example, the two
    916916            Chinese characters <code class="code">䜠奜</code> are represented as two
     
    926926            may then be completed by applying parallel deletion and inverse transposition of the
    927927            UTF-16 bitstreams\cite{Cameron2008}. </p>
    928 <p id="idp456704"> Rather than immediately paying the costs of deletion and transposition just for
     928<p id="idp473408"> Rather than immediately paying the costs of deletion and transposition just for
    929929            transcoding, however, icXML defers these steps so that the deletion masks for several
    930930            stages of processing may be combined. In particular, this includes core XML requirements
     
    935935            first marking CR positions, performing two bit parallel operations to transform the
    936936            marked CRs into LFs, and then marking for deletion any LF that is found immediately
    937             after the marked CR as shown by the Pablo source code in Figure
    938             \ref{fig:LBnormalization}.
    939               <div class="figure" id="idp458944">
     937            after the marked CR as shown by the Pablo source code in
     938              <a class="xref" href="#fig-LBnormalization">Figure 3</a>.
     939              <div class="figure" id="fig-LBnormalization">
    940940<p class="title">Figure 3</p>
    941941<div class="figure-contents">
    942942<div class="caption">Line Break Normalization Logic</div>
    943 <pre class="programlisting" id="idp459632">
     943<pre class="programlisting" id="idp477488">
    944944# XML 1.0 line-break normalization rules.
    945945if lex.CR:
     
    957957</div>
    958958         </p>
    959 <p id="idp461008"> In essence, the deletion masks for transcoding and for line break normalization each
     959<p id="idp478832"> In essence, the deletion masks for transcoding and for line break normalization each
    960960            represent a bitwise filter; these filters can be combined using bitwise-or so that the
    961961            parallel deletion algorithm need only be applied once. </p>
    962 <p id="idp461664"> A further application of combined filtering is the processing of XML character and
     962<p id="idp479488"> A further application of combined filtering is the processing of XML character and
    963963            entity references. Consider, for example, the references <code class="code">&amp;</code> or
    964964               <code class="code">&lt;</code>. which must be replaced in XML processing with the single
     
    973973            UTF-16 code unit. In the case, that this is not true, it is addressed in
    974974            post-processing. </p>
    975 <p id="idp466480"> The final step of combined filtering occurs during the process of reducing markup
     975<p id="idp484304"> The final step of combined filtering occurs during the process of reducing markup
    976976            data to tag bytes preceding each significant XML transition as described in
    977977            section~\ref{section:arch:contentstream}. Overall, icXML avoids separate buffer copying
     
    983983            Haswell architecture. </p>
    984984</div>
    985 <div class="section" id="idp467808">
     985<div class="section" id="idp485632">
    986986<h3 class="title" style="clear: both">Content Stream</h3>
    987 <p id="idp468480"> A relatively-unique concept for icXML is the use of a filtered content stream.
     987<p id="idp486304"> A relatively-unique concept for icXML is the use of a filtered content stream.
    988988            Rather that parsing an XML document in its original format, the input is transformed
    989989            into one that is easier for the parser to iterate through and produce the sequential
     
    993993           
    994994            through the parallel filtering algorithm, described in section \ref{sec:parfilter}. </p>
    995 <p id="idp470992"> Combined with the symbol stream, the parser traverses the content stream to
     995<p id="idp488816"> Combined with the symbol stream, the parser traverses the content stream to
    996996            effectively reconstructs the input document in its output form. The initial <span class="ital">0</span> indicates an empty content string. The following
    997997               <code class="code">&gt;</code> indicates that a start tag without any attributes is the first
     
    10051005            null character in the content stream in parallel, which in turn means the parser can
    10061006            directly jump to the end of every string without scanning for it. </p>
    1007 <p id="idp474384"> Following <code class="code">'fee'</code> is a <code class="code">=</code>, which marks the
     1007<p id="idp492208"> Following <code class="code">'fee'</code> is a <code class="code">=</code>, which marks the
    10081008            existence of an attribute. Because all of the intra-element was performed in the Parabix
    10091009            Subsystem, this must be a legal attribute. Since attributes can only occur within start
     
    10191019            that the appropriate scope-nesting rules have been applied. </p>
    10201020</div>
    1021 <div class="section" id="idp478528">
     1021<div class="section" id="idp496352">
    10221022<h3 class="title" style="clear: both">Namespace Handling</h3>
    1023 <p id="idp479616"> In XML, namespaces prevents naming conflicts when multiple vocabularies are used
     1023<p id="idp497440"> In XML, namespaces prevents naming conflicts when multiple vocabularies are used
    10241024            together. It is especially important when a vocabulary application-dependant meaning,
    10251025            such as when XML or SVG documents are embedded within XHTML files. Namespaces are bound
     
    10381038            uniquely-named items because the current vocabulary is determined by the namespace(s)
    10391039            that are in-scope. </p>
    1040 <div class="table-wrapper" id="idp486784">
     1040<div class="table-wrapper" id="idp504560">
    10411041<p class="title">Table V</p>
    1042 <div class="caption"><p id="idp487296">XML Namespace Example</p></div>
     1042<div class="caption"><p id="idp505072">XML Namespace Example</p></div>
    10431043<table class="table">
    10441044<colgroup span="1">
     
    10741074</table>
    10751075</div>
    1076 <p id="idp496272"> In both Xerces and icXML, every URI has a one-to-one mapping to a URI ID. These
     1076<p id="idp514048"> In both Xerces and icXML, every URI has a one-to-one mapping to a URI ID. These
    10771077            persist for the lifetime of the application through the use of a global URI pool. Xerces
    10781078            maintains a stack of namespace scopes that is pushed (popped) every time a start tag
     
    10821082            those that declare a set of namespaces upfront and never change them, and (2) those that
    10831083            repeatedly modify the namespaces in predictable patterns. </p>
    1084 <p id="idp497408"> For that reason, icXML contains an independent namespace stack and utilizes bit
     1084<p id="idp515184"> For that reason, icXML contains an independent namespace stack and utilizes bit
    10851085            vectors to cheaply perform
    10861086             When a prefix is
     
    10961096            found using a bit-scan intrinsic. A namespace binding table, similar to Table
    10971097            \ref{tbl:namespace1}, provides the actual URI ID. </p>
    1098 <div class="table-wrapper" id="idp501872">
     1098<div class="table-wrapper" id="idp519648">
    10991099<p class="title">Table VI</p>
    1100 <div class="caption"><p id="idp502384">Namespace Binding Table Example</p></div>
     1100<div class="caption"><p id="idp520160">Namespace Binding Table Example</p></div>
    11011101<table class="table">
    11021102<colgroup span="1">
     
    11391139</table>
    11401140</div>
    1141 <p id="idp518624">
     1141<p id="idp536240">
    11421142           
    11431143           
     
    11451145           
    11461146         </p>
    1147 <p id="idp520528"> To ensure that scoping rules are adhered to, whenever a start tag is encountered,
     1147<p id="idp538224"> To ensure that scoping rules are adhered to, whenever a start tag is encountered,
    11481148            any modification to the currently visible namespaces is calculated and stored within a
    11491149            stack of bit vectors denoting the locally modified namespace bindings. When an end tag
     
    11541154         </p>
    11551155</div>
    1156 <div class="section" id="idp521984">
     1156<div class="section" id="idp539680">
    11571157<h3 class="title" style="clear: both">Error Handling</h3>
    1158 <p id="idp522656">
     1158<p id="idp540352">
    11591159           
    11601160            Xerces outputs error messages in two ways: through the programmer API and as thrown
     
    11651165            \ref{fig:icxml-arch}, icXML is divided into two sections: the Parabix Subsystem and
    11661166            Markup Processor, each with its own system for detecting and producing error messages. </p>
    1167 <p id="idp524288"> Within the Parabix Subsystem, all computations are performed in parallel, a block at
     1167<p id="idp541984"> Within the Parabix Subsystem, all computations are performed in parallel, a block at
    11681168            a time. Errors are derived as artifacts of bitstream calculations, with a 1-bit marking
    11691169            the byte-position of an error within a block, and the type of error is determined by the
     
    11981198            detected, the sum of those skipped positions is subtracted from the distance to
    11991199            determine the actual column number. </p>
    1200 <p id="idp529776"> The Markup Processor is a state-driven machine. As such, error detection within it
     1200<p id="idp547472"> The Markup Processor is a state-driven machine. As such, error detection within it
    12011201            is very similar to Xerces. However, reporting the correct line/column is a much more
    12021202            difficult problem. The Markup Processor parses the content stream, which is a series of
     
    12121212</div>
    12131213</div>
    1214 <div class="section" id="idp532208">
     1214<div class="section" id="idp549968">
    12151215<h2 class="title" style="clear: both">Multithreading with Pipeline Parallelism</h2>
    1216 <p id="idp532912"> As discussed in section \ref{background:xerces}, Xerces can be considered a FSM
     1216<p id="idp550608"> As discussed in section \ref{background:xerces}, Xerces can be considered a FSM
    12171217         application. These are "embarrassingly
    12181218         sequential."\cite{Asanovic:EECS-2006-183} and notoriously difficult to
     
    12221222         well into the general model of pipeline parallelism, in which each thread is in charge of a
    12231223         single module or group of modules. </p>
    1224 <p id="idp534768"> The most straightforward division of work in icXML is to separate the Parabix Subsystem
     1224<p id="idp552464"> The most straightforward division of work in icXML is to separate the Parabix Subsystem
    12251225         and the Markup Processor into distinct logical layers into two separate stages. The
    12261226         resultant application, <span class="ital">icXML-p</span>, is a course-grained
     
    12431243            <code class="code">T<sub>2</sub></code> to finish reading the shared data before it can
    12441244         reuse the memory space. </p>
    1245 <p id="idp544496">
     1245<p id="idp561584">
    12461246        <div class="figure" id="threads_timeline1">
    12471247<p class="title">Figure 4: Thread Balance in Two-Stage Pipelines</p>
    12481248<div class="figure-contents">
    1249 <div class="mediaobject" id="idp545936"><img alt="png image (threads_timeline1.png)" src="threads_timeline1.png" width="500cm"></div>
    1250 <div class="mediaobject" id="idp547712"><img alt="png image (threads_timeline2.png)" src="threads_timeline2.png" width="500cm"></div>
     1249<div class="mediaobject" id="idp562976"><img alt="png image (threads_timeline1.png)" src="threads_timeline1.png" width="500cm"></div>
     1250<div class="mediaobject" id="idp564752"><img alt="png image (threads_timeline2.png)" src="threads_timeline2.png" width="500cm"></div>
    12511251<div class="caption"></div>
    12521252</div>
    12531253</div>
    12541254      </p>
    1255 <p id="idp550160"> Overall, our design is intended to benefit a range of applications. Conceptually, we
     1255<p id="idp567200"> Overall, our design is intended to benefit a range of applications. Conceptually, we
    12561256         consider two design points. The first, the parsing performed by the Parabix Subsystem
    12571257         dominates at 67% of the overall cost, with the cost of application processing (including
     
    12591259         scenario, the cost of application processing dominates at 60%, while the cost of XML
    12601260         parsing represents an overhead of 40%. </p>
    1261 <p id="idp551072"> Our design is predicated on a goal of using the Parabix framework to achieve a 50% to
     1261<p id="idp568112"> Our design is predicated on a goal of using the Parabix framework to achieve a 50% to
    12621262         100% improvement in the parsing engine itself. In a best case scenario, a 100% improvement
    12631263         of the Parabix Subsystem for the design point in which XML parsing dominates at 67% of the
     
    12671267         about 33% of the original work. In this case, Amdahl's law predicts that we could expect up
    12681268         to a 3x speedup at best. </p>
    1269 <p id="idp552192"> At the other extreme of our design range, we consider an application in which core
     1269<p id="idp569232"> At the other extreme of our design range, we consider an application in which core
    12701270         parsing cost is 40%. Assuming the 2x speedup of the Parabix Subsystem over the
    12711271         corresponding Xerces core, single-threaded icXML delivers a 25% speedup. However, the most
     
    12731273         the entire latency of parsing within the serial time required by the application. In this
    12741274         case, we achieve an overall speedup in processing time by 1.67x. </p>
    1275 <p id="idp553136"> Although the structure of the Parabix Subsystem allows division of the work into
     1275<p id="idp570176"> Although the structure of the Parabix Subsystem allows division of the work into
    12761276         several pipeline stages and has been demonstrated to be effective for four pipeline stages
    12771277         in a research prototype \cite{HPCA2012}, our analysis here suggests that the further
     
    12811281         the cost of application logic that could match reductions in core parsing cost. </p>
    12821282</div>
    1283 <div class="section" id="idp554320">
     1283<div class="section" id="idp572016">
    12841284<h2 class="title" style="clear: both">Performance</h2>
    1285 <p id="idp554992"> We evaluate Xerces-C++ 3.1.1, icXML, icXML-p against two benchmarking applications: the
     1285<p id="idp572688"> We evaluate Xerces-C++ 3.1.1, icXML, icXML-p against two benchmarking applications: the
    12861286         Xerces C++ SAXCount sample application, and a real world GML to SVG transformation
    12871287         application. We investigated XML parser performance using an Intel Core i7 quad-core (Sandy
     
    12891289         L1 cache, 256 kB (per core) L2 cache, 8 MB L3 cache) running the 64-bit version of Ubuntu
    12901290         12.04 (Linux). </p>
    1291 <p id="idp555904"> We analyzed the execution profiles of each XML parser using the performance counters
     1291<p id="idp573600"> We analyzed the execution profiles of each XML parser using the performance counters
    12921292         found in the processor. We chose several key hardware events that provide insight into the
    12931293         profile of each application and indicate if the processor is doing useful work. The set of
     
    12971297         collection of hardware performance monitoring statistics. In addition, we used the Linux
    12981298         perf \cite{perf} utility to collect per core hardware events. </p>
    1299 <div class="section" id="idp557040">
     1299<div class="section" id="idp574736">
    13001300<h3 class="title" style="clear: both">Xerces C++ SAXCount</h3>
    1301 <p id="idp557712"> Xerces comes with sample applications that demonstrate salient features of the
     1301<p id="idp575408"> Xerces comes with sample applications that demonstrate salient features of the
    13021302            parser. SAXCount is the simplest such application: it counts the elements, attributes
    13031303            and characters of a given XML file using the (event based) SAX API and prints out the
    13041304            totals. </p>
    1305 <p id="idp558416"> Table \ref{XMLDocChars} shows the document characteristics of the XML input files
     1305<p id="idp576112"> Table \ref{XMLDocChars} shows the document characteristics of the XML input files
    13061306            selected for the Xerces C++ SAXCount benchmark. The jaw.xml represents document-oriented
    13071307            XML inputs and contains the three-byte and four-byte UTF-8 sequence required for the
    13081308            UTF-8 encoding of Japanese characters. The remaining data files are data-oriented XML
    13091309            documents and consist entirely of single byte encoded ASCII characters.
    1310   <div class="table-wrapper" id="idp559152">
     1310  <div class="table-wrapper" id="idp576848">
    13111311<p class="title">Table VII</p>
    1312 <div class="caption"><p id="idp559664">XML Document Characteristics</p></div>
     1312<div class="caption"><p id="idp577360">XML Document Characteristics</p></div>
    13131313<table class="table">
    13141314<colgroup span="1">
     
    13591359</div>           
    13601360</p>
    1361 <p id="idp575296"> A key predictor of the overall parsing performance of an XML file is markup
     1361<p id="idp592944"> A key predictor of the overall parsing performance of an XML file is markup
    13621362            density\footnote{ Markup Density: the ratio of markup bytes used to define the structure
    13631363            of the document vs. its file size.}. This metric has substantial influence on the
     
    13661366            of document-oriented and data-oriented XML files to analyze performance over a spectrum
    13671367            of markup densities. </p>
    1368 <p id="idp576304"> Figure \ref{perf_SAX} compares the performance of Xerces, icXML and pipelined icXML
     1368<p id="idp593952"> Figure \ref{perf_SAX} compares the performance of Xerces, icXML and pipelined icXML
    13691369            in terms of CPU cycles per byte for the SAXCount application. The speedup for icXML over
    13701370            Xerces is 1.3x to 1.8x. With two threads on the multicore machine, icXML-p can achieve
     
    13731373            icXML-p performs better as markup-density increases because the work performed by each
    13741374            stage is well balanced in this application. </p>
    1375 <p id="idp577344">
     1375<p id="idp594992">
    13761376        <div class="figure" id="perf_SAX">
    13771377<p class="title">Figure 5: SAXCount Performance Comparison</p>
    13781378<div class="figure-contents">
    1379 <div class="mediaobject" id="idp578736"><img alt="png image (perf_SAX.png)" src="perf_SAX.png" width="500cm"></div>
     1379<div class="mediaobject" id="idp596336"><img alt="png image (perf_SAX.png)" src="perf_SAX.png" width="500cm"></div>
    13801380<div class="caption"></div>
    13811381</div>
     
    13831383         </p>
    13841384</div>
    1385 <div class="section" id="idp581280">
     1385<div class="section" id="idp598880">
    13861386<h3 class="title" style="clear: both">GML2SVG</h3>
    1387 <p id="idp581952">       As a more substantial application of XML processing, the GML-to-SVG (GML2SVG) application
     1387<p id="idp599552">       As a more substantial application of XML processing, the GML-to-SVG (GML2SVG) application
    13881388was chosen.   This application transforms geospatially encoded data represented using
    13891389an XML representation in the form of Geography Markup Language (GML) \cite{lake2004geography}
     
    13971397a known XML format for the purpose of analysis and restructuring to meet
    13981398the requirements of an alternative format.</p>
    1399 <p id="idp583280">Our GML to SVG data translations are executed on GML source data
     1399<p id="idp600880">Our GML to SVG data translations are executed on GML source data
    14001400modelling the city of Vancouver, British Columbia, Canada.
    14011401The GML source document set
     
    14071407<p class="title">Figure 6: Performance Comparison for GML2SVG</p>
    14081408<div class="figure-contents">
    1409 <div class="mediaobject" id="idp585312"><img alt="png image (Throughput.png)" src="Throughput.png" width="500cm"></div>
     1409<div class="mediaobject" id="idp602864"><img alt="png image (Throughput.png)" src="Throughput.png" width="500cm"></div>
    14101410<div class="caption"></div>
    14111411</div>
    14121412</div>
    1413 <p id="idp587600">Figure \ref{perf_GML2SVG} compares the performance of the GML2SVG application linked against
     1413<p id="idp605152">Figure \ref{perf_GML2SVG} compares the performance of the GML2SVG application linked against
    14141414the Xerces, icXML and icXML-p.   
    14151415On the GML workload with this application, single-thread icXML
     
    14181418Using icXML-p, a further throughput increase to 111 MB/sec was recorded,
    14191419approximately a 2X speedup.</p>
    1420 <p id="idp588416">An important aspect of icXML is the replacement of much branch-laden
     1420<p id="idp605968">An important aspect of icXML is the replacement of much branch-laden
    14211421sequential code inside Xerces with straight-line SIMD code using far
    14221422fewer branches.  Figure \ref{branchmiss_GML2SVG} shows the corresponding
     
    14291429<p class="title">Figure 7: Comparative Branch Misprediction Rate</p>
    14301430<div class="figure-contents">
    1431 <div class="mediaobject" id="idp590528"><img alt="png image (BM.png)" src="BM.png" width="500cm"></div>
     1431<div class="mediaobject" id="idp30880"><img alt="png image (BM.png)" src="BM.png" width="500cm"></div>
    14321432<div class="caption"></div>
    14331433</div>
    14341434</div>
    1435 <p id="idp592816">The behaviour of the three versions with respect to L1 cache misses per kB is shown
     1435<p id="idp33168">The behaviour of the three versions with respect to L1 cache misses per kB is shown
    14361436in Figure \ref{cachemiss_GML2SVG}.   Improvements are shown in both instruction-
    14371437and data-cache performance with the improvements in instruction-cache
     
    14451445<p class="title">Figure 8: Comparative Cache Miss Rate</p>
    14461446<div class="figure-contents">
    1447 <div class="mediaobject" id="idp594944"><img alt="png image (CM.png)" src="CM.png" width="500cm"></div>
     1447<div class="mediaobject" id="idp35296"><img alt="png image (CM.png)" src="CM.png" width="500cm"></div>
    14481448<div class="caption"></div>
    14491449</div>
    14501450</div>
    1451 <p id="idp597232">One caveat with this study is that the GML2SVG application did not exhibit
     1451<p id="idp37584">One caveat with this study is that the GML2SVG application did not exhibit
    14521452a relative balance of processing between application code and Xerces library
    14531453code reaching the 33% figure.  This suggests that for this application and
     
    14571457</div>
    14581458</div>
    1459 <div class="section" id="idp598304">
     1459<div class="section" id="idp624032">
    14601460<h2 class="title" style="clear: both">Conclusion and Future Work</h2>
    1461 <p id="idp598992"> This paper is the first case study documenting the significant performance benefits
     1461<p id="idp624720"> This paper is the first case study documenting the significant performance benefits
    14621462         that may be realized through the integration of parallel bitstream technology into existing
    14631463         widely-used software libraries. In the case of the Xerces-C++ XML parser, the combined
     
    14691469         technologies, this is an important case study demonstrating the general feasibility of
    14701470         these techniques. </p>
    1471 <p id="idp600272"> The further development of icXML to move beyond 2-stage pipeline parallelism is
     1471<p id="idp626000"> The further development of icXML to move beyond 2-stage pipeline parallelism is
    14721472         ongoing, with realistic prospects for four reasonably balanced stages within the library.
    14731473         For applications such as GML2SVG which are dominated by time spent on XML parsing, such a
    14741474         multistage pipelined parsing library should offer substantial benefits. </p>
    1475 <p id="idp601040"> The example of XML parsing may be considered prototypical of finite-state machines
     1475<p id="idp626768"> The example of XML parsing may be considered prototypical of finite-state machines
    14761476         applications which have sometimes been considered "embarassingly
    14771477         sequential" and so difficult to parallelize that "nothing
     
    14791479         point in making the case that parallelization can indeed be helpful across a broad array of
    14801480         application types. </p>
    1481 <p id="idp602416"> To overcome the software engineering challenges in applying parallel bitstream
     1481<p id="idp628144"> To overcome the software engineering challenges in applying parallel bitstream
    14821482         technology to existing software systems, it is clear that better library and tool support
    14831483         is needed. The techniques used in the implementation of icXML and documented in this paper
     
    14861486      </p>
    14871487</div>
    1488 <div class="bibliography" id="idp603904">
     1488<div class="bibliography" id="idp629600">
    14891489<h2 class="title" style="clear:both">Bibliography</h2>
    14901490<p class="bibliomixed" id="XMLChip09">[Leventhal and Lemoine 2009] Leventhal, Michael and
  • docs/Balisage13/Bal2013came0601/Bal2013came0601.xml

    r3054 r3055  
    167167            state. This introduces implicit dependencies between the various tasks within the
    168168            application that make it difficult to optimize for performance. As a complex software
    169             system, no one feature dominates the overall parsing performance. Table I
     169              system, no one feature dominates the overall parsing performance. <xref linkend="xerces-profile"/>
    170170            shows the execution time profile of the top ten functions in a
    171171            typical run. Even if it were possible, Amdahl's Law dictates that tackling any one of
     
    177177            expected that a comprehensive restructuring is required, involving all aspects of the
    178178            parser. </para>
    179              <table>
     179             <table xml:id="xerces-profile">
    180180                  <caption>
    181181                     <para>Execution Time of Top 10 Xerces Functions</para>
     
    207207            data from long blocks of input data by using one register bit per single input byte. To
    208208            facilitate this, the input data is first transposed into a set of basis bit streams.
    209               For example Table II shows  the ASCII bytes for the string "<code>b7&lt;A</code>" with
    210                 the corresponding  8 basis bit streams, b<subscript>0</subscript> through  b<subscript>7</subscript> shown in Table III.
     209              For example, <xref linkend="xml-bytes"/> shows  the ASCII bytes for the string "<code>b7&lt;A</code>" with
     210                the corresponding  8 basis bit streams, b<subscript>0</subscript> through  b<subscript>7</subscript> shown in  <xref linkend="xml-bits"/>.
    211211-->
    212212            <!-- The bits used to construct $\tt <subscript>7</subscript>$ have been highlighted in this example. -->
     
    230230            <!-- the classification cost could be amortized over many character classes.--> multiple
    231231            classes can share the classification cost. </para>
    232          <table>
     232         <table xml:id="xml-bytes">
    233233                  <caption>
    234234                     <para>XML Source Data</para>
     
    248248 
    249249</table>         
    250          <table>
     250         <table xml:id="xml-bits">
    251251                  <caption>
    252252                     <para>8-bit ASCII Basis Bit Streams</para>
     
    278278         <!-- process, intra-element well-formedness validation is performed on each block -->
    279279         <!-- of text. -->
    280          <para> Consider, for example, the XML source data stream shown in the first line of Table IV.
     280         <para> Consider, for example, the XML source data stream shown in the first line of <xref linkend="derived"/>.
    281281The remaining lines of this figure show
    282282            several parallel bit streams that are computed in Parabix-style parsing, with each bit
     
    290290            (using the technique of bitstream addition \cite{cameron-EuroPar2011}), namely streams
    291291            marking the element names, attribute names and attribute values of tags. </para>
    292             <table>
     292            <table xml:id="derived">
    293293                  <caption>
    294294                     <para>XML Source Data and Derived Parallel Bit Streams</para>
     
    395395        </figure>
    396396         <para> In icXML functions are grouped into logical components. As shown in Figure
    397             \ref{fig:icxml-arch}, two major categories exist: (1) the Parabix Subsystem and (2) the
     397             <xref linkend="xerces-arch"/>, two major categories exist: (1) the Parabix Subsystem and (2) the
    398398            Markup Processor. All tasks in (1) use the Parabix Framework \cite{HPCA2012}, which
    399399            represents data as a set of parallel bitstreams. The <emphasis role="ital">Character Set
     
    519519            first marking CR positions, performing two bit parallel operations to transform the
    520520            marked CRs into LFs, and then marking for deletion any LF that is found immediately
    521             after the marked CR as shown by the Pablo source code in Figure
    522               <xref  linkend="fig:LBnormalization">.
    523               <figure id="fig:LBnormalization">
     521            after the marked CR as shown by the Pablo source code in
     522              <xref  linkend="fig-LBnormalization"/>.
     523              <figure xml:id="fig-LBnormalization">
    524524                <caption>Line Break Normalization Logic</caption>
    525525  <programlisting>
Note: See TracChangeset for help on using the changeset viewer.