Changeset 1415 for docs


Ignore:
Timestamp:
Aug 31, 2011, 7:13:21 PM (8 years ago)
Author:
ashriram
Message:

buf fixes up to 7

Location:
docs/HPCA2012
Files:
3 edited

Legend:

Unmodified
Added
Removed
  • docs/HPCA2012/04-methodology.tex

    r1407 r1415  
     1\pagebreak
    12\section{Evaluation Framework}
    23\label{section:methodology}
     
    5556SSE extensions have been available on commodity Intel processors for
    5657over a decade since the Pentium III. They have steadily evolved with
    57 improvements in instruction latency, cache interface, and register
    58 resources, and the addition domain specific instructions. Here we
     58improvements in instruction latency, cache interface, register
     59resources, and the addition of domain specific instructions. Here we
    5960investigate SIMD extensions across three different generations of
    60 intel processors. Table \ref{hwinfo} describes the Intel multicores we
    61 investigate. We compare the energy and performance profile of the
    62 Parabix under the platforms.  We also analyze the implementation
    63 specifics of SIMD extensions under various microarchitecture. We we
    64 evalute both the legacy SSE and newer AVX extensions supported by
     61intel processors (hardware details in Table \ref{hwinfo}). We compare
     62the energy and performance profile of the Parabix under the platforms.
     63We also analyze the implementation specifics of SIMD extensions under
     64various microarchitectures and the newer AVX extensions supported by
    6565Sandybridge.
    6666
     67
    6768We investigated the execution profiles of each XML parser
    68 using the Performance Monitoring Counter (PMC) found in the processor.
     69using the performance counters found in the processor.
    6970We chose several key hardware events that provide insight into the profile of each
    7071application and indicate if the processor is doing useful work
  • docs/HPCA2012/05-corei3.tex

    r1411 r1415  
    5050\subsection{Branch Mispredictions}
    5151\label{section:XML-branches}
    52 In general, performance is limited by branch mispredictions.
    53 Unfortunately, it is difficult to reduce the branch misprediction rate of
    54 traditional XML parsers due to:
    55 (1) the variable length nature of the syntactic elements contained within XML documents;
    56 (2) a data dependent characteristic, and
    57 (3) the extensive set of syntax constraints imposed by the XML 1.0/1.1 specifications.
     52It is hard to handle branch mispredictions in traditional
     53XML parsers due to: (1) the variable length nature of the syntactic
     54elements contained within XML documents; (2) a data dependent
     55characteristic, and (3) the extensive set of syntax constraints
     56imposed by the XML 1.0/1.1 specifications.
    5857% Branch mispredictions are known
    5958% to signficantly degrade XML parsing performance in proportion to the markup density of the source document
Note: See TracChangeset for help on using the changeset viewer.