[jira] [Commented] (LUCENE-5352) PackedInts.fastestFormatAndBits returns slower format for Direct

["Adrien Grand (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/LUCENE-5352?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13835460#comment-13835460 ] 

Adrien Grand commented on LUCENE-5352:
--------------------------------------

I ran tests as well and got similar results.

It may be possible to optimize this reader (maybe reversing the bytes in the call to readLong takes substantial time on little-endian computers (just guessing)?) but taking a step back, I'm wondering that maybe this aligned format should be deprecated. As [~toke] noted last year, the in-memory impl is not faster than Packed64 on all CPUs, so if the speed-up is not clear for disk-based structures either, I think we should just deprecate it?

> PackedInts.fastestFormatAndBits returns slower format for Direct
> ----------------------------------------------------------------
>
>                 Key: LUCENE-5352
>                 URL: https://issues.apache.org/jira/browse/LUCENE-5352
>             Project: Lucene - Core
>          Issue Type: Bug
>            Reporter: Robert Muir
>
> from some simple testing, DirectPacked64SingleBlockReader seems slower than DirectPackedReader (at least in a few cases i tested, more testing needed, but look at the code).
> this is unintuitive, because by passing e.g. FASTER you are making it slower.



--
This message was sent by Atlassian JIRA
(v6.1#6144)

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