lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mark Miller (JIRA)" <j...@apache.org>
Subject [jira] Commented: (LUCENE-1997) Explore performance of multi-PQ vs single-PQ sorting API
Date Mon, 26 Oct 2009 22:34:59 GMT

    [ https://issues.apache.org/jira/browse/LUCENE-1997?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12770242#action_12770242
] 

Mark Miller commented on LUCENE-1997:
-------------------------------------

Waiting for Mike's changes to try and do some different varying tests, but while I wait, here
is a 1.6 test to go with that 1.5 test:

JAVA:
java version "1.6.0_15"
Java(TM) SE Runtime Environment (build 1.6.0_15-b03)
Java HotSpot(TM) 64-Bit Server VM (build 14.1-b02, mixed mode)

OS:
Linux quad-laptop 2.6.31-14-generic #48-Ubuntu SMP x86_64 GNU/Linux

||Source||Seg size||Query||Tot hits||Sort||Top N||QPS old||QPS new||Pct change||
|wiki|log|1|317914|title|10|105.49|106.68|{color:green}1.1%{color}|
|wiki|log|1|317914|title|25|108.78|110.37|{color:green}1.5%{color}|
|wiki|log|1|317914|title|50|106.79|106.35|{color:red}-0.4%{color}|
|wiki|log|1|317914|title|100|105.26|100.10|{color:red}-4.9%{color}|
|wiki|log|1|317914|title|500|91.03|72.09|{color:red}-20.8%{color}|
|wiki|log|1|317914|title|1000|80.17|53.61|{color:red}-33.1%{color}|
|wiki|log|<all>|1000000|title|10|114.44|116.63|{color:green}1.9%{color}|
|wiki|log|<all>|1000000|title|25|113.03|116.77|{color:green}3.3%{color}|
|wiki|log|<all>|1000000|title|50|111.52|112.33|{color:green}0.7%{color}|
|wiki|log|<all>|1000000|title|100|86.97|101.80|{color:green}17.1%{color}|
|wiki|log|<all>|1000000|title|500|77.79|66.45|{color:red}-14.6%{color}|
|wiki|log|<all>|1000000|title|1000|70.76|46.45|{color:red}-34.4%{color}|
|random|log|<all>|1000000|rand string|10|118.47|112.77|{color:red}-4.8%{color}|
|random|log|<all>|1000000|rand string|25|117.98|109.12|{color:red}-7.5%{color}|
|random|log|<all>|1000000|rand string|50|117.51|105.05|{color:red}-10.6%{color}|
|random|log|<all>|1000000|rand string|100|86.44|96.08|{color:green}11.2%{color}|
|random|log|<all>|1000000|rand string|500|77.03|60.82|{color:red}-21.0%{color}|
|random|log|<all>|1000000|rand string|1000|68.84|40.33|{color:red}-41.4%{color}|
|random|log|<all>|1000000|country|10|118.25|110.85|{color:red}-6.3%{color}|
|random|log|<all>|1000000|country|25|117.55|110.41|{color:red}-6.1%{color}|
|random|log|<all>|1000000|country|50|117.22|103.35|{color:red}-11.8%{color}|
|random|log|<all>|1000000|country|100|87.01|94.02|{color:green}8.1%{color}|
|random|log|<all>|1000000|country|500|79.53|60.63|{color:red}-23.8%{color}|
|random|log|<all>|1000000|country|1000|72.62|40.15|{color:red}-44.7%{color}|
|random|log|<all>|1000000|rand int|10|125.54|116.18|{color:red}-7.5%{color}|
|random|log|<all>|1000000|rand int|25|124.83|111.81|{color:red}-10.4%{color}|
|random|log|<all>|1000000|rand int|50|124.17|108.26|{color:red}-12.8%{color}|
|random|log|<all>|1000000|rand int|100|88.58|100.63|{color:green}13.6%{color}|
|random|log|<all>|1000000|rand int|500|81.81|64.23|{color:red}-21.5%{color}|
|random|log|<all>|1000000|rand int|1000|74.14|42.79|{color:red}-42.3%{color}|


> Explore performance of multi-PQ vs single-PQ sorting API
> --------------------------------------------------------
>
>                 Key: LUCENE-1997
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1997
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: Search
>    Affects Versions: 2.9
>            Reporter: Michael McCandless
>            Assignee: Michael McCandless
>         Attachments: LUCENE-1997.patch, LUCENE-1997.patch, LUCENE-1997.patch, LUCENE-1997.patch
>
>
> Spinoff from recent "lucene 2.9 sorting algorithm" thread on java-dev,
> where a simpler (non-segment-based) comparator API is proposed that
> gathers results into multiple PQs (one per segment) and then merges
> them in the end.
> I started from John's multi-PQ code and worked it into
> contrib/benchmark so that we could run perf tests.  Then I generified
> the Python script I use for running search benchmarks (in
> contrib/benchmark/sortBench.py).
> The script first creates indexes with 1M docs (based on
> SortableSingleDocSource, and based on wikipedia, if available).  Then
> it runs various combinations:
>   * Index with 20 balanced segments vs index with the "normal" log
>     segment size
>   * Queries with different numbers of hits (only for wikipedia index)
>   * Different top N
>   * Different sorts (by title, for wikipedia, and by random string,
>     random int, and country for the random index)
> For each test, 7 search rounds are run and the best QPS is kept.  The
> script runs singlePQ then multiPQ, and records the resulting best QPS
> for each and produces table (in Jira format) as output.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


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


Mime
View raw message