[
https://issues.apache.org/jira/browse/OPENJPA-160?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12476351
]
John Stecher commented on OPENJPA-160:
--------------------------------------
All performance throughput numbers are run without the profiler on a 2-way
Linux/Intel machine. 1410 tps at 99% CPU without the patch, 1413 tps at 99%
CPU with both patches. Those numbers are average throughput over a 3 minute
run with another 3 minute run as a warmup.
Flipping the profiler on only after gathering those numbers is what shows the
above graphs.
Using Jprofiler as well as IBM internal tools.
I am in agreement with you that I am a little taken aback by the overhead of
the newinstance call but I can tell you that the <init> you see in the profile
is the constructor and only the constructor my knowledge of the profiling
tools, everything else under newinstance is the class being setup by the VM and
the class members being initialized.
Pooling the objects and thus eliminating the object creation completely leads
to a tps rate of 1506 using the same methodology talked about above.
> Reuse BrokerImpl objects
> ------------------------
>
> Key: OPENJPA-160
> URL: https://issues.apache.org/jira/browse/OPENJPA-160
> Project: OpenJPA
> Issue Type: Sub-task
> Reporter: Michael Dick
> Assigned To: Patrick Linskey
> Attachments: newprofile.jpg, openjpa-160-patch.txt,
> openjpa-160-patch.txt, perf2.jpg, perf3.jpg
>
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
