lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Michael McCandless (JIRA)" <j...@apache.org>
Subject [jira] Commented: (LUCENE-1997) Explore performance of multi-PQ vs single-PQ sorting API
Date Tue, 27 Oct 2009 17:12:59 GMT

    [ https://issues.apache.org/jira/browse/LUCENE-1997?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12770578#action_12770578
] 

Michael McCandless commented on LUCENE-1997:
--------------------------------------------

OK yet more results, this time on 5M doc indexes:

JAVA:
java version "1.6.0_14"
Java(TM) SE Runtime Environment (build 1.6.0_14-b08)
Java HotSpot(TM) 64-Bit Server VM (build 14.0-b16, mixed mode)


OS:
SunOS rhumba 5.11 snv_111b i86pc i386 i86pc Solaris


||Source||Seg size||Query||Tot hits||Sort||Top N||QPS old||QPS new||Pct change||
|wiki|log|1|1169569|title|10|26.80|28.08|{color:green}4.8%{color}|
|wiki|log|1|1169569|title|25|26.52|27.68|{color:green}4.4%{color}|
|wiki|log|1|1169569|title|50|26.37|26.58|{color:green}0.8%{color}|
|wiki|log|1|1169569|title|100|25.91|26.50|{color:green}2.3%{color}|
|wiki|log|1|1169569|title|500|24.28|23.78|{color:red}-2.1%{color}|
|wiki|log|1|1169569|title|1000|22.64|21.32|{color:red}-5.8%{color}|
|wiki|log|2|1088167|title|10|28.23|29.16|{color:green}3.3%{color}|
|wiki|log|2|1088167|title|25|27.90|28.54|{color:green}2.3%{color}|
|wiki|log|2|1088167|title|50|28.04|28.62|{color:green}2.1%{color}|
|wiki|log|2|1088167|title|100|27.57|27.88|{color:green}1.1%{color}|
|wiki|log|2|1088167|title|500|25.85|25.27|{color:red}-2.2%{color}|
|wiki|log|2|1088167|title|1000|23.65|22.33|{color:red}-5.6%{color}|
|wiki|log|<all>|5000000|title|10|27.32|30.47|{color:green}11.5%{color}|
|wiki|log|<all>|5000000|title|25|26.29|31.16|{color:green}18.5%{color}|
|wiki|log|<all>|5000000|title|50|26.64|30.03|{color:green}12.7%{color}|
|wiki|log|<all>|5000000|title|100|21.37|29.67|{color:green}38.8%{color}|
|wiki|log|<all>|5000000|title|500|21.05|26.15|{color:green}24.2%{color}|
|wiki|log|<all>|5000000|title|1000|19.83|21.97|{color:green}10.8%{color}|
|random|log|<all>|5000000|rand string|10|29.00|32.39|{color:green}11.7%{color}|
|random|log|<all>|5000000|rand string|25|28.39|32.36|{color:green}14.0%{color}|
|random|log|<all>|5000000|rand string|50|28.74|31.90|{color:green}11.0%{color}|
|random|log|<all>|5000000|rand string|100|28.44|31.16|{color:green}9.6%{color}|
|random|log|<all>|5000000|rand string|500|28.20|28.03|{color:red}-0.6%{color}|
|random|log|<all>|5000000|rand string|1000|19.47|24.50|{color:green}25.8%{color}|
|random|log|<all>|5000000|country|10|29.18|32.15|{color:green}10.2%{color}|
|random|log|<all>|5000000|country|25|28.66|31.74|{color:green}10.7%{color}|
|random|log|<all>|5000000|country|50|28.02|32.00|{color:green}14.2%{color}|
|random|log|<all>|5000000|country|100|28.68|31.27|{color:green}9.0%{color}|
|random|log|<all>|5000000|country|500|28.16|26.66|{color:red}-5.3%{color}|
|random|log|<all>|5000000|country|1000|19.58|23.03|{color:green}17.6%{color}|
|random|log|<all>|5000000|rand int|10|28.75|29.51|{color:green}2.6%{color}|
|random|log|<all>|5000000|rand int|25|28.74|28.96|{color:green}0.8%{color}|
|random|log|<all>|5000000|rand int|50|29.29|28.63|{color:red}-2.3%{color}|
|random|log|<all>|5000000|rand int|100|29.14|28.38|{color:red}-2.6%{color}|
|random|log|<all>|5000000|rand int|500|28.44|24.59|{color:red}-13.5%{color}|
|random|log|<all>|5000000|rand int|1000|20.06|21.71|{color:green}8.2%{color}|
|wiki|balanced|1|1169558|title|10|25.70|27.06|{color:green}5.3%{color}|
|wiki|balanced|1|1169558|title|25|25.91|27.26|{color:green}5.2%{color}|
|wiki|balanced|1|1169558|title|50|25.84|26.63|{color:green}3.1%{color}|
|wiki|balanced|1|1169558|title|100|25.51|25.66|{color:green}0.6%{color}|
|wiki|balanced|1|1169558|title|500|24.39|22.32|{color:red}-8.5%{color}|
|wiki|balanced|1|1169558|title|1000|22.66|18.74|{color:red}-17.3%{color}|
|wiki|balanced|2|1088162|title|10|27.86|28.77|{color:green}3.3%{color}|
|wiki|balanced|2|1088162|title|25|27.46|28.37|{color:green}3.3%{color}|
|wiki|balanced|2|1088162|title|50|27.49|28.05|{color:green}2.0%{color}|
|wiki|balanced|2|1088162|title|100|27.64|26.25|{color:red}-5.0%{color}|
|wiki|balanced|2|1088162|title|500|25.52|23.26|{color:red}-8.9%{color}|
|wiki|balanced|2|1088162|title|1000|23.77|19.69|{color:red}-17.2%{color}|
|wiki|balanced|<all>|5000000|title|10|27.13|31.05|{color:green}14.4%{color}|
|wiki|balanced|<all>|5000000|title|25|27.81|30.47|{color:green}9.6%{color}|
|wiki|balanced|<all>|5000000|title|50|27.99|29.81|{color:green}6.5%{color}|
|wiki|balanced|<all>|5000000|title|100|22.00|29.35|{color:green}33.4%{color}|
|wiki|balanced|<all>|5000000|title|500|21.03|24.12|{color:green}14.7%{color}|
|wiki|balanced|<all>|5000000|title|1000|20.43|19.80|{color:red}-3.1%{color}|
|random|balanced|<all>|5000000|rand string|10|28.16|32.13|{color:green}14.1%{color}|
|random|balanced|<all>|5000000|rand string|25|28.40|31.44|{color:green}10.7%{color}|
|random|balanced|<all>|5000000|rand string|50|28.13|31.07|{color:green}10.5%{color}|
|random|balanced|<all>|5000000|rand string|100|27.93|31.25|{color:green}11.9%{color}|
|random|balanced|<all>|5000000|rand string|500|27.49|26.91|{color:red}-2.1%{color}|
|random|balanced|<all>|5000000|rand string|1000|19.13|22.69|{color:green}18.6%{color}|
|random|balanced|<all>|5000000|country|10|27.74|32.11|{color:green}15.8%{color}|
|random|balanced|<all>|5000000|country|25|28.31|32.15|{color:green}13.6%{color}|
|random|balanced|<all>|5000000|country|50|27.98|31.17|{color:green}11.4%{color}|
|random|balanced|<all>|5000000|country|100|27.66|30.85|{color:green}11.5%{color}|
|random|balanced|<all>|5000000|country|500|27.57|26.25|{color:red}-4.8%{color}|
|random|balanced|<all>|5000000|country|1000|19.01|21.27|{color:green}11.9%{color}|
|random|balanced|<all>|5000000|rand int|10|28.16|28.57|{color:green}1.5%{color}|
|random|balanced|<all>|5000000|rand int|25|28.39|28.56|{color:green}0.6%{color}|
|random|balanced|<all>|5000000|rand int|50|28.08|27.84|{color:red}-0.9%{color}|
|random|balanced|<all>|5000000|rand int|100|27.81|27.23|{color:red}-2.1%{color}|
|random|balanced|<all>|5000000|rand int|500|27.80|24.01|{color:red}-13.6%{color}|
|random|balanced|<all>|5000000|rand int|1000|19.87|20.14|{color:green}1.4%{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,
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