[
https://issues.apache.org/jira/browse/LUCENE-6198?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14320076#comment-14320076
]
Adrien Grand commented on LUCENE-6198:
--------------------------------------
In order to make sure that I did not miss anything, I updated Robert's patch to
a recent trunk and benchmarked it against my previous patch (the one that does
not flatten in ConjunctionDISI) and they perform the same (baseline is Robert's
patch, patch is the previous patch):
{noformat}
TaskQPS baseline StdDev QPS patch StdDev
Pct diff
AndMedPhraseLowTerm 213.42 (10.6%) 207.08 (14.3%)
-3.0% ( -25% - 24%)
AndHighPhraseMedTerm 18.06 (2.8%) 17.95 (2.8%)
-0.6% ( -6% - 5%)
AndHighPhraseLowTerm 44.37 (4.4%) 44.31 (4.2%)
-0.1% ( -8% - 8%)
AndHighPhraseHighTerm 11.47 (3.4%) 11.47 (3.5%)
0.1% ( -6% - 7%)
AndMedPhraseMedTerm 47.85 (6.1%) 48.28 (5.8%)
0.9% ( -10% - 13%)
AndMedPhraseHighTerm 19.58 (3.2%) 19.79 (3.2%)
1.1% ( -5% - 7%)
{noformat}
So I think we're good?
> two phase intersection
> ----------------------
>
> Key: LUCENE-6198
> URL: https://issues.apache.org/jira/browse/LUCENE-6198
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Robert Muir
> Attachments: LUCENE-6198.patch, LUCENE-6198.patch, LUCENE-6198.patch,
> LUCENE-6198.patch, LUCENE-6198.patch, phrase_intersections.tasks
>
>
> Currently some scorers have to do a lot of per-document work to determine if
> a document is a match. The simplest example is a phrase scorer, but there are
> others (spans, sloppy phrase, geospatial, etc).
> Imagine a conjunction with two MUST clauses, one that is a term that matches
> all odd documents, another that is a phrase matching all even documents.
> Today this conjunction will be very expensive, because the zig-zag
> intersection is reading a ton of useless positions.
> The same problem happens with filteredQuery and anything else that acts like
> a conjunction.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
