[ 
https://issues.apache.org/jira/browse/LUCENE-3664?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13174703#comment-13174703
 ] 

Simon Willnauer commented on LUCENE-3664:
-----------------------------------------

here are some benchmark results using Java 1.6.0_30

{code}

                Task   QPS trunkStdDev trunkQPS jit_patchStdDev jit_patch      
Pct diff
        SloppyPhrase        2.73        0.16        2.72        0.15  -10% -   
11%
      TermBGroup1M1P       43.89        0.53       44.53        0.50    0% -    
3%
        TermBGroup1M       30.12        0.59       30.66        0.28   -1% -    
4%
         TermGroup1M       25.77        0.63       26.26        0.46   -2% -    
6%
                Term       49.82        2.63       51.23        2.15   -6% -   
13%
              Fuzzy1       73.03        3.70       75.19        1.73   -4% -   
10%
              Fuzzy2       48.70        2.29       50.23        0.97   -3% -   
10%
             Respell       51.05        3.14       52.75        1.51   -5% -   
13%
         AndHighHigh       11.18        0.36       11.56        0.21   -1% -    
8%
           OrHighMed       10.70        0.32       11.12        0.26   -1% -    
9%
          OrHighHigh        6.19        0.21        6.44        0.17   -2% -   
10%
             Prefix3       53.55        1.84       55.73        1.03   -1% -    
9%
            Wildcard       22.75        0.49       23.81        0.41    0% -    
8%
            SpanNear        5.48        0.37        5.79        0.13   -3% -   
15%
            PKLookup       96.23        2.02      101.66        4.39    0% -   
12%
              Phrase        6.55        0.61        6.93        0.41   -9% -   
23%
          AndHighMed       22.81        1.05       24.14        0.38    0% -   
12%
              IntNRQ        5.75        0.52        6.27        0.38   -6% -   
27%
{code}
                
> Speed up SegementDocsAndPositionsEnum by making it more friendly for JIT 
> optimizations
> --------------------------------------------------------------------------------------
>
>                 Key: LUCENE-3664
>                 URL: https://issues.apache.org/jira/browse/LUCENE-3664
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: core/search
>    Affects Versions: 4.0
>            Reporter: Simon Willnauer
>            Priority: Minor
>             Fix For: 4.0
>
>         Attachments: LUCENE-3664.patch
>
>
> LUCENE-3648 applied some refactoring to make SegmentDocsEnum reuse some code 
> and divorce the liveDocs and no-liveDocs case into sep classes to make more 
> friendly for jit optimizations. I did the same thing for 
> SegmentDocsAndPositions[AndPayloads]Enum removing a couple of hundred lines 
> of code abstracting it into a base class. patch follows in a sec...

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: 
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to