[
https://issues.apache.org/jira/browse/UIMA-3638?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Marshall Schor updated UIMA-3638:
---------------------------------
Description: In large UIMA deployments with multiple threads, some
synchronization contention is detected in RsFullFeatNames and RsLang, and the
caching that occurs with MetaDataObject_impl. Replace with concurrentHashMap,
and for RsFullFeatNames - discard the weakpointers to keys, which is not much
of a hazard - because the tables are relatively small vs other memory uses.
(was: In large UIMA deployments with multiple threads, some synchronization
contention is detected in RsFullFeatNames and RsLang. Replace with
concurrentHashMap, and for RsFullFeatNames - discard the weakpointers to keys,
which is not much of a hazard - because the tables are relatively small vs
other memory uses.)
> Remove potential sync bottleneck in RsFullFeatNames and RsLang
> --------------------------------------------------------------
>
> Key: UIMA-3638
> URL: https://issues.apache.org/jira/browse/UIMA-3638
> Project: UIMA
> Issue Type: Improvement
> Components: Core Java Framework
> Affects Versions: 2.5.0SDK
> Reporter: Marshall Schor
> Assignee: Marshall Schor
> Priority: Minor
> Fix For: 2.5.1SDK
>
>
> In large UIMA deployments with multiple threads, some synchronization
> contention is detected in RsFullFeatNames and RsLang, and the caching that
> occurs with MetaDataObject_impl. Replace with concurrentHashMap, and for
> RsFullFeatNames - discard the weakpointers to keys, which is not much of a
> hazard - because the tables are relatively small vs other memory uses.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
