[
https://issues.apache.org/jira/browse/LUCENE-1990?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Toke Eskildsen updated LUCENE-1990:
-----------------------------------
Attachment: generated_performance-te20100226.txt
LUCENE-1990-te20100226c.patch
{quote}
Airplane blocking snow drifts!? Where on earth are you anyway?
{quote}
In Denmark. The guy responsible for clearing the runway did indeed clear the
runway. He just forgot that the plane needs to taxi into the runway in the
first place. That made us miss our connecting flight.
{quote}
It's very interesting that align is never a win -- I think in that case
removing it makes sense? It'll be a nice simplification.
{quote}
Well, practically never wins for the machines I tested on and never wins with
my implementation.
{quote}
Can you use @lucene.internal instead of the NOTE that I had put on the classes?
{quote}
Done... I think. I'm not very good at this part, so if someone else wants to do
some cleanup i JavaDoc and such, they are very welcome by me.
{quote}
We lost "final" in the RamUsageEstimator constants
{quote}
Strange. Oh well, fixed.
{quote}
Did we ever test performance of the specialized (generated) decoders using
switch statements?
{quote}
I just did a quick hack in order to measure performance and I was very
surprised that the generated switch-based implementations performs so well.
It's nearly on par with packed most of the time and exceeds it in some cases. I
only tested on 3 machines though. The hack is in the
LUCENE-1990-te20100226c.patch and is called when the performance test is
executed.
Attachment generated_performance-te20100226.txt contains measurements where the
py-generated code is tested together with the other implementations.
Note to self: Switch is not equivalent to a series of if-else, when we're
talking performance and when we switch without omissions in the cases.
> Add unsigned packed int impls in oal.util
> -----------------------------------------
>
> Key: LUCENE-1990
> URL: https://issues.apache.org/jira/browse/LUCENE-1990
> Project: Lucene - Java
> Issue Type: Improvement
> Components: Index
> Reporter: Michael McCandless
> Priority: Minor
> Attachments: generated_performance-te20100226.txt,
> LUCENE-1990-te20100122.patch, LUCENE-1990-te20100210.patch,
> LUCENE-1990-te20100212.patch, LUCENE-1990-te20100223.patch,
> LUCENE-1990-te20100226.patch, LUCENE-1990-te20100226b.patch,
> LUCENE-1990-te20100226c.patch, LUCENE-1990.patch,
> LUCENE-1990_PerformanceMeasurements20100104.zip, performance-te20100226.txt
>
>
> There are various places in Lucene that could take advantage of an
> efficient packed unsigned int/long impl. EG the terms dict index in
> the standard codec in LUCENE-1458 could subsantially reduce it's RAM
> usage. FieldCache.StringIndex could as well. And I think "load into
> RAM" codecs like the one in TestExternalCodecs could use this too.
> I'm picturing something very basic like:
> {code}
> interface PackedUnsignedLongs {
> long get(long index);
> void set(long index, long value);
> }
> {code}
> Plus maybe an iterator for getting and maybe also for setting. If it
> helps, most of the usages of this inside Lucene will be "write once"
> so eg the set could make that an assumption/requirement.
> And a factory somewhere:
> {code}
> PackedUnsignedLongs create(int count, long maxValue);
> {code}
> I think we should simply autogen the code (we can start from the
> autogen code in LUCENE-1410), or, if there is an good existing impl
> that has a compatible license that'd be great.
> I don't have time near-term to do this... so if anyone has the itch,
> please jump!
--
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
