[
https://issues.apache.org/jira/browse/LUCENE-1997?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12771017#action_12771017
]
Jake Mannix commented on LUCENE-1997:
-------------------------------------
But part of the point is that you don't have to get the values - you can have a
fast in-memory structure which just encodes their sort-order, right? This is
the whole point of using the ordinal - you pre-sort all of the possible values
to get the ordinals, and now arbitrarily complex comparator reduces to int
compare at sort time(*). In the custom comparators we use, for example, this
allows for even sorting by multivalued fields in a custom way, via the simple
compare(int doca, int docb) way.
(*) This reminds me: Mike, you switched the compare for ord values from being
"return ordA - ordB" to being "return ordA > ordB ? 1 : (ordA == ordB ? 0 :
-1)", on the basis of int overflow at some point, right? This is only true if
we're really sorting by integers, which could overflow - if they're ordinals,
then these are both non-negative numbers, and their difference will always be
greater than -MAX_INT, so the branching can be avoided in this innermost
comparison in this case.
> Explore performance of multi-PQ vs single-PQ sorting API
> --------------------------------------------------------
>
> Key: LUCENE-1997
> URL: https://issues.apache.org/jira/browse/LUCENE-1997
> Project: Lucene - Java
> Issue Type: Improvement
> Components: Search
> Affects Versions: 2.9
> Reporter: Michael McCandless
> Assignee: Michael McCandless
> Attachments: LUCENE-1997.patch, LUCENE-1997.patch, LUCENE-1997.patch,
> LUCENE-1997.patch, LUCENE-1997.patch, LUCENE-1997.patch, LUCENE-1997.patch,
> LUCENE-1997.patch
>
>
> Spinoff from recent "lucene 2.9 sorting algorithm" thread on java-dev,
> where a simpler (non-segment-based) comparator API is proposed that
> gathers results into multiple PQs (one per segment) and then merges
> them in the end.
> I started from John's multi-PQ code and worked it into
> contrib/benchmark so that we could run perf tests. Then I generified
> the Python script I use for running search benchmarks (in
> contrib/benchmark/sortBench.py).
> The script first creates indexes with 1M docs (based on
> SortableSingleDocSource, and based on wikipedia, if available). Then
> it runs various combinations:
> * Index with 20 balanced segments vs index with the "normal" log
> segment size
> * Queries with different numbers of hits (only for wikipedia index)
> * Different top N
> * Different sorts (by title, for wikipedia, and by random string,
> random int, and country for the random index)
> For each test, 7 search rounds are run and the best QPS is kept. The
> script runs singlePQ then multiPQ, and records the resulting best QPS
> for each and produces table (in Jira format) as output.
--
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
