Changeset 1031 for docs/PACT2011


Ignore:
Timestamp:
Mar 25, 2011, 7:22:10 PM (8 years ago)
Author:
cameron
Message:

Meth clean-up

Location:
docs/PACT2011
Files:
2 edited

Legend:

Unmodified
Added
Removed
  • docs/PACT2011/03-research.tex

    r1003 r1031  
    11\section{Parabix}
    2 \label{section:reserach}
     2\label{section:parabix}
    33%Describe key technology behind Parabix
    44%Introduce SIMD;
     
    4747\begin{center}
    4848\begin{tabular}{cr}\\
    49 source data $\vartriangleright$ & \verb`<t1>abc</t1><t2/>`\\
     49source data & \verb`<t1>abc</t1><t2/>`\\
    5050$B_0$ & \verb`..1.1.1.1.1....1.`\\
    5151$B_1$ & \verb`...1.11.1..1..111`\\
     
    6868\begin{center}
    6969\begin{tabular}{lr}\\
    70 source data $\vartriangleright$ & \verb`<t1>abc<t1/><t2/>`\\
     70source data & \verb`<t1>abc<t1/><t2/>`\\
    7171% $N = $ name chars & \verb`.11.111.11...11..`\\
    7272$M_0 = 1$ & \verb`1................`\\
     
    107107%In Parabix2, we replace the sequential single-cursor parsing using bit scan instructions with a parallel parsing method using bitstream addition. Unlike the single cursor approach of Parabix1 and conceptually of traditional sequential approach, in Parabix2 multiple cursors positions are processed in parallel.
    108108
    109 In Parabix2, we replace the sequential single-cursor parsing using bit scan instructions with a parallel parsing method using bitstream addition. Unlike the single-cursor approach of Parabix1 (and conceptually of all sequential XML parsers), Parabix2 processes multiple cursors in parallel. For example, using the source data from Figure \ref{fig:Parabix1StarttagExample}, Figure \ref{fig:Parabix2StarttagExample} shows how Parabix2 identifies and moves each of the start tag markers forwards to the corresponding end tag. Like Parabix1, we assume that $N$ (the name chars) has been computed using the basis bitstreams and that
     109In Parabix2, we replace the sequential single-cursor parsing using bit scan
     110instructions with a parallel parsing method using bitstream addition.
     111Unlike the single-cursor approach of Parabix1 (and conceptually of all sequential XML parsers),
     112Parabix2 processes multiple cursors in parallel. For example, using the source data from
     113Figure \ref{fig:Parabix1StarttagExample}, Figure \ref{fig:Parabix2StarttagExample} shows how
     114Parabix2 identifies and moves each of the start tag markers forwards to the corresponding end tag.
     115%Like Parabix1, we assume that $N$ (the name chars) has been computed using the basis bitstreams and that
    110116
    111117\begin{figure}[h]
    112118\begin{center}
    113119\begin{tabular}{lr}\\
    114 source data $\vartriangleright$ & \verb`<t1>abc<t1/><t2/>`\\
     120source data & \verb`<t1>abc<t1/><t2/>`\\
    115121$N = $ name chars & \verb`.11.111.11...11..`\\
    116122$M_0 = [<]$ & \verb`1......1....1....`\\
     
    124130\end{figure}
    125131
    126 In general, the set of bit positions in a marker bitstream may be considered to be the current parsing positions of multiple parses taking place in parallel throughout the source data stream. A further aspect of the parallel method is that conditional branch statements used to identify syntax error at each each parsing position are eliminated. Although we do not show it in the prior examples, error bitstreams can be used to identify any well-formedness errors found during the parsing process. Error positions are gathered and processed in as a final post processing step. Hence, Parabix2 offers additional parallelism over Parabix1 in the form of multiple cursor parsing as well as significanlty reduces branch misprediction penalty.
     132In general, the set of bit positions in a marker bitstream may be considered to be the current parsing
     133positions of multiple parses taking place in parallel throughout the source data stream.
     134A further aspect of the parallel method is that conditional branch statements used to identify
     135syntax error at each each parsing position are eliminated. Although we do not show it in the prior examples,
     136error bitstreams can be used to identify any well-formedness errors found during the parsing process. Error positions are gathered and
     137processed in as a final post processing step. Hence, Parabix2 offers additional parallelism over Parabix1 in the form of multiple cursor
     138parsing as well as further reducing branch misprediction penalties.
    127139
    128140
  • docs/PACT2011/04-methodology.tex

    r1030 r1031  
    44In this section, we describe our methodology for the measurements and
    55investigation of XML parsing energy consumption and performance.  In
    6 brief, for each of the XML parsers under study we measure
     6brief, for each of the XML parsers under study we propose to measure
    77and evaluate the energy consumption required to carry out XML
    88well-formedness checking, under a variety of workloads, and as
    9 executed on three different Intel processors.
     9executed on three different Intel cores.
    1010
    11 To begin our study, we investigate each of the XML
     11To begin our study, we propose to first investigate each of the XML
    1212parsers in terms of the PMCs hardware events as listed in the
    1313following subsection. Based on the recommendation of previous
     
    7777\subsection{Workloads}\label{workloads}
    7878
    79 Documented-oriented and data-oriented XML are characterized by lower and higher markup density respectively.  Markup density is defined
     79Markup density is defined
    8080as the ratio of the total markup contained within an XML file to the
    8181total XML document size.  This metric may have substantial influence
    82 on the performance of XML parsing.  As such we choose workloads with a
    83 spectrum of markup densities.
     82on the performance of XML parsing. 
     83We use a mixture of document-oriented and data-oriented XML
     84files in our study to provide workloads with a spectrum of
     85markup densities.
    8486
    8587Table \ref{XMLDocChars} shows the document characteristics of the XML
     
    8890containing three-byte and four-byte UTF8 sequence.  The remaining
    8991files are data-oriented inputs and consist of only ASCII
    90 characters.\cite{CameronHerdyLin2008}
     92characters. 
    9193
    9294
     
    164166\subsection{PMC Hardware Events}\label{events}
    165167
    166 Each of the hardware events selected relates to the energy consumption
    167 due to one or more hardware units. For example, total branch miss
    168 predictions corresponds to the use of the branch misprediction unit.
     168Each of the hardware events selected relates to performance
     169and energy features associated with
     170one or more hardware units.   For example, total branch mispredictions
     171relate to the branch predictor and branch target buffer capacity.
    169172
    170 Initial PMC hardware event set:
     173The set of PMC events used include the following.
    171174\begin{itemize}
    172175\item Processor Cycles
     
    182185clamp applied on the 12V wires that supply power to the processor
    183186sockets. The clamp detects the magnetic field created by the flowing
    184 current and converts it into voltage lev- els (1mV per 1A
     187current and converts it into voltage levels (1mV per 1A
    185188current). The voltage levels are then monitored by an Agilent 34410a
    186 multimeter at the granu- larity of 100 samples per second. This
    187 measurement cap- tures the power to the processor package, including
     189multimeter at the granularity of 100 samples per second. This
     190measurement captures the power to the processor package, including
    188191cores, caches, Northbridge memory controller, and the quick-path
    189 interconnects. \cite{clamp}.
     192interconnects \cite{clamp}.
Note: See TracChangeset for help on using the changeset viewer.