lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Michael McCandless (JIRA)" <>
Subject [jira] [Updated] (LUCENE-7563) BKD index should compress unused leading bytes
Date Fri, 02 Dec 2016 11:42:59 GMT


Michael McCandless updated LUCENE-7563:
    Attachment: LUCENE-7563.patch

New patch; I think it's ready.

This breaks out a private BKD implementation for {{SimpleText}} which
is a nice cleanup for the core BKD implementation, e.g. {{BKDReader}}
is now final; its strange protected constructor is gone; protected
methods are now private.

This patch also implements [~jpountz]'s last compression idea, to often
use only 1 byte to encode prefix, splitDim and first-byte-delta of the
suffix instead of the 2 bytes required in the previous iterations.
This gives another ~4-5% further compression improvement:

  * sparse-sorted -> 2.37 MB

  * sparse -> 2.07 MB

  * dense -> 2.00 MB

And the OpenStreetMaps geo benchmark:

  * geo3d -> 1.75 MB

  * LatLonPoint -> 1.72 MB

I'm running the 2B BKD and Points tests now ... if those pass, I plan
to push to master first and let this bake a bit before backporting.

> BKD index should compress unused leading bytes
> ----------------------------------------------
>                 Key: LUCENE-7563
>                 URL:
>             Project: Lucene - Core
>          Issue Type: Improvement
>            Reporter: Michael McCandless
>             Fix For: master (7.0), 6.4
>         Attachments: LUCENE-7563.patch, LUCENE-7563.patch, LUCENE-7563.patch, LUCENE-7563.patch
> Today the BKD (points) in-heap index always uses {{dimensionNumBytes}} per dimension,
but if e.g. you are indexing {{LongPoint}} yet only use the bottom two bytes in a given segment,
we shouldn't store all those leading 0s in the index.

This message was sent by Atlassian JIRA

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message