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 Wed, 28 Oct 2009 10:23:59 GMT

    [ https://issues.apache.org/jira/browse/LUCENE-1997?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12770878#action_12770878
] 

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

OK results using last patch:

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|balanced|1|1169558|title|10|25.87|25.36|{color:red}-2.0%{color}|
|wiki|balanced|1|1169558|title|25|24.95|26.51|{color:green}6.3%{color}|
|wiki|balanced|1|1169558|title|50|25.27|25.28|{color:green}0.0%{color}|
|wiki|balanced|1|1169558|title|100|24.81|24.25|{color:red}-2.3%{color}|
|wiki|balanced|1|1169558|title|500|22.91|21.37|{color:red}-6.7%{color}|
|wiki|balanced|1|1169558|title|1000|21.45|17.73|{color:red}-17.3%{color}|
|wiki|balanced|2|1088162|title|10|27.45|26.97|{color:red}-1.7%{color}|
|wiki|balanced|2|1088162|title|25|26.85|26.88|{color:green}0.1%{color}|
|wiki|balanced|2|1088162|title|50|26.66|26.66|{color:green}0.0%{color}|
|wiki|balanced|2|1088162|title|100|26.45|26.27|{color:red}-0.7%{color}|
|wiki|balanced|2|1088162|title|500|23.53|21.99|{color:red}-6.5%{color}|
|wiki|balanced|2|1088162|title|1000|23.03|19.38|{color:red}-15.8%{color}|
|wiki|balanced|<all>|5000000|title|10|26.75|27.41|{color:green}2.5%{color}|
|wiki|balanced|<all>|5000000|title|25|26.18|27.74|{color:green}6.0%{color}|
|wiki|balanced|<all>|5000000|title|50|24.28|27.06|{color:green}11.4%{color}|
|wiki|balanced|<all>|5000000|title|100|20.82|26.96|{color:green}29.5%{color}|
|wiki|balanced|<all>|5000000|title|500|19.71|22.41|{color:green}13.7%{color}|
|wiki|balanced|<all>|5000000|title|1000|19.65|18.18|{color:red}-7.5%{color}|
|random|balanced|<all>|5000000|rand string|10|25.75|27.66|{color:green}7.4%{color}|
|random|balanced|<all>|5000000|rand string|25|24.32|27.59|{color:green}13.4%{color}|
|random|balanced|<all>|5000000|rand string|50|21.42|27.19|{color:green}26.9%{color}|
|random|balanced|<all>|5000000|rand string|100|19.30|26.66|{color:green}38.1%{color}|
|random|balanced|<all>|5000000|rand string|500|17.80|22.00|{color:green}23.6%{color}|
|random|balanced|<all>|5000000|rand string|1000|17.30|18.12|{color:green}4.7%{color}|
|random|balanced|<all>|5000000|country|10|28.63|28.04|{color:red}-2.1%{color}|
|random|balanced|<all>|5000000|country|25|25.02|27.57|{color:green}10.2%{color}|
|random|balanced|<all>|5000000|country|50|20.50|26.81|{color:green}30.8%{color}|
|random|balanced|<all>|5000000|country|100|19.95|26.34|{color:green}32.0%{color}|
|random|balanced|<all>|5000000|country|500|18.43|22.19|{color:green}20.4%{color}|
|random|balanced|<all>|5000000|country|1000|18.69|17.54|{color:red}-6.2%{color}|
|random|balanced|<all>|5000000|rand int|10|28.39|29.46|{color:green}3.8%{color}|
|random|balanced|<all>|5000000|rand int|25|26.97|27.73|{color:green}2.8%{color}|
|random|balanced|<all>|5000000|rand int|50|29.76|27.87|{color:red}-6.4%{color}|
|random|balanced|<all>|5000000|rand int|100|28.63|26.59|{color:red}-7.1%{color}|
|random|balanced|<all>|5000000|rand int|500|27.18|21.67|{color:red}-20.3%{color}|
|random|balanced|<all>|5000000|rand int|1000|25.79|16.92|{color:red}-34.4%{color}|
|wiki|log|1|1169569|title|10|23.98|23.32|{color:red}-2.8%{color}|
|wiki|log|1|1169569|title|25|25.55|26.53|{color:green}3.8%{color}|
|wiki|log|1|1169569|title|50|25.44|26.04|{color:green}2.4%{color}|
|wiki|log|1|1169569|title|100|24.43|25.92|{color:green}6.1%{color}|
|wiki|log|1|1169569|title|500|23.61|22.77|{color:red}-3.6%{color}|
|wiki|log|1|1169569|title|1000|21.88|20.48|{color:red}-6.4%{color}|
|wiki|log|2|1088167|title|10|28.03|27.48|{color:red}-2.0%{color}|
|wiki|log|2|1088167|title|25|27.20|27.57|{color:green}1.4%{color}|
|wiki|log|2|1088167|title|50|27.75|27.59|{color:red}-0.6%{color}|
|wiki|log|2|1088167|title|100|26.77|26.69|{color:red}-0.3%{color}|
|wiki|log|2|1088167|title|500|25.11|24.21|{color:red}-3.6%{color}|
|wiki|log|2|1088167|title|1000|23.27|21.75|{color:red}-6.5%{color}|
|wiki|log|<all>|5000000|title|10|26.51|28.24|{color:green}6.5%{color}|
|wiki|log|<all>|5000000|title|25|25.22|27.58|{color:green}9.4%{color}|
|wiki|log|<all>|5000000|title|50|24.51|28.05|{color:green}14.4%{color}|
|wiki|log|<all>|5000000|title|100|20.64|27.47|{color:green}33.1%{color}|
|wiki|log|<all>|5000000|title|500|19.69|24.32|{color:green}23.5%{color}|
|wiki|log|<all>|5000000|title|1000|19.30|20.56|{color:green}6.5%{color}|
|random|log|<all>|5000000|rand string|10|25.57|26.94|{color:green}5.4%{color}|
|random|log|<all>|5000000|rand string|25|25.56|28.12|{color:green}10.0%{color}|
|random|log|<all>|5000000|rand string|50|21.33|27.63|{color:green}29.5%{color}|
|random|log|<all>|5000000|rand string|100|18.96|27.31|{color:green}44.0%{color}|
|random|log|<all>|5000000|rand string|500|18.74|23.72|{color:green}26.6%{color}|
|random|log|<all>|5000000|rand string|1000|16.88|20.39|{color:green}20.8%{color}|
|random|log|<all>|5000000|country|10|28.96|26.71|{color:red}-7.8%{color}|
|random|log|<all>|5000000|country|25|24.91|27.41|{color:green}10.0%{color}|
|random|log|<all>|5000000|country|50|19.81|27.19|{color:green}37.3%{color}|
|random|log|<all>|5000000|country|100|20.80|26.80|{color:green}28.8%{color}|
|random|log|<all>|5000000|country|500|18.90|23.63|{color:green}25.0%{color}|
|random|log|<all>|5000000|country|1000|18.52|19.31|{color:green}4.3%{color}|
|random|log|<all>|5000000|rand int|10|27.48|28.12|{color:green}2.3%{color}|
|random|log|<all>|5000000|rand int|25|28.81|27.76|{color:red}-3.6%{color}|
|random|log|<all>|5000000|rand int|50|29.60|28.41|{color:red}-4.0%{color}|
|random|log|<all>|5000000|rand int|100|29.16|26.44|{color:red}-9.3%{color}|
|random|log|<all>|5000000|rand int|500|27.81|21.96|{color:red}-21.0%{color}|
|random|log|<all>|5000000|rand int|1000|25.61|18.48|{color:red}-27.8%{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, 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