[ 
https://issues.apache.org/jira/browse/LUCENE-1410?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12852930#action_12852930
 ] 

Renaud Delbru commented on LUCENE-1410:
---------------------------------------

{quote}
The fact that S9 is already doing better than VInt is encouraging. Since (P)FOR 
can do even better than S9, when using (P)FOR only for longer sequences, I'd 
expect a real performance boost for queries using frequently occurring terms in 
the index.
{quote}

So, you're suggesting that maybe in my benchmark, the decoded sequence are not 
long enough to see the performance improvements of FOR/PFOR ? 

Also, in my benchmark, VINT means block-based VInt. So, there is the same 
overhead, i.e., decompress the full block even if a partial sequence is needed, 
than for FOR, PFOR, S9 and the others. But, even with these settings, as you 
see, the average query time is smaller when using VInt than FOR/PFOR. 

{quote}
Also, I'd recommend to verify query results for each method. S9 as I 
implemented it is only tested by its own test cases.
{quote}

We implemented more unit tests for each codec. Also, in the benchmark, I output 
the number of hits for each query per codec, and the number of hits for each 
codec is the same (but I haven't checked if they returns all the same document 
ids).

> PFOR implementation
> -------------------
>
>                 Key: LUCENE-1410
>                 URL: https://issues.apache.org/jira/browse/LUCENE-1410
>             Project: Lucene - Java
>          Issue Type: New Feature
>          Components: Other
>            Reporter: Paul Elschot
>            Priority: Minor
>         Attachments: autogen.tgz, for-summary.txt, 
> LUCENE-1410-codecs.tar.bz2, LUCENE-1410b.patch, LUCENE-1410c.patch, 
> LUCENE-1410d.patch, LUCENE-1410e.patch, TermQueryTests.tgz, TestPFor2.java, 
> TestPFor2.java, TestPFor2.java
>
>   Original Estimate: 21840h
>  Remaining Estimate: 21840h
>
> Implementation of Patched Frame of Reference.

-- 
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-unsubscr...@lucene.apache.org
For additional commands, e-mail: java-dev-h...@lucene.apache.org

Reply via email to