[jira] [Commented] (LUCENE-5236) Use broadword bit selection in EliasFanoDecoder

["Sebastiano Vigna (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/LUCENE-5236?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13930217#comment-13930217 ] 

Sebastiano Vigna commented on LUCENE-5236:
------------------------------------------

Sorry guys—just happened to read at this thread. My 2€¢:

- Rank9/Select9 and full ranking structures, not rank/select-in-a-word algorithms.
- I'm using Long.bitCount(), Long.numberOfTralilingZeros() etc. everywhere since Jan 2013. Intrinsification is a bet, but it pays, and using them will convince them IBM and others to intrinsify. If only Oracle would add Long.clearLowestBitSet() ...
- The alternative algorithm for sideways addition (a.k.a. popcount) should be called "broadword" or Wilkes–Wheeler–Gill.
- Selection gets better and better every day. http://vigna.di.unimi.it/select.php tries to keep track of the improvement. The current code in it.unimi.dsi.bits.Fast is the best I'm aware of.

Ciao!

> Use broadword bit selection in EliasFanoDecoder
> -----------------------------------------------
>
>                 Key: LUCENE-5236
>                 URL: https://issues.apache.org/jira/browse/LUCENE-5236
>             Project: Lucene - Core
>          Issue Type: Improvement
>            Reporter: Paul Elschot
>            Assignee: Adrien Grand
>            Priority: Minor
>             Fix For: 4.6
>
>         Attachments: LUCENE-5236.patch, LUCENE-5236.patch, LUCENE-5236.patch, LUCENE-5236.patch, TestDocIdSetBenchmark.java
>
>
> Try and speed up decoding



--
This message was sent by Atlassian JIRA
(v6.2#6252)

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