lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Adrien Grand (JIRA)" <>
Subject [jira] [Commented] (LUCENE-4171) Performance improvements to Packed64
Date Thu, 28 Jun 2012 12:57:43 GMT


Adrien Grand commented on LUCENE-4171:

Even if {{Packed64}} used one or two additional blocks to avoid branches, the reader and the
writer always read/wrote the exact number of required blocks (this is why ctor #3 reads {{size}}
longs instead of {{blocks.length == size + 1}} blocks). So this should not be a problem.

Can you fix ctor #1 so that it uses the size method instead of {{valueCount * bitsPerValue
/ BLOCK_SIZE + 1}} (this may create one more block than required: for example, if bitsPerValue=1
and valueCount=64, blocks will have a length of 2 while only 1 block is required)?
> Performance improvements to Packed64
> ------------------------------------
>                 Key: LUCENE-4171
>                 URL:
>             Project: Lucene - Java
>          Issue Type: Sub-task
>          Components: core/other
>    Affects Versions: 4.0, 5.0
>         Environment: Tested with 4 different Intel machines
>            Reporter: Toke Eskildsen
>            Priority: Minor
>              Labels: performance
>         Attachments: LUCENE-4171.patch, LUCENE-4171.patch, LUCENE-4171.patch
>   Original Estimate: 4h
>  Remaining Estimate: 4h
> Based on the performance measurements of PackedInts.Mutable's in LUCENE-4062, a new version
of Packed64 has been created that is consistently faster than the old Packed64 for both get
and set.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:!default.jspa
For more information on JIRA, see:


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

View raw message