[
https://issues.apache.org/jira/browse/LOG4J2-744?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Scott Harrington updated LOG4J2-744:
------------------------------------
Attachment: ClockBenchmark-jdk1.7.0_45.txt
Summary of results for Java 7 (full JMH output attached). Looks like the
'instanceof' downcast adds about 10-20ns to a 85-125ns Log4jLogEvent
constructor. Ignore CachedClock which seems to suffer from thread contention (I
configured this JMH run to use 4 threads).
CentOS 6.5 on E5-2667 v2 @ 3.30GHz
{noformat}
Benchmark
Mode Samples Mean Mean error Units
o.a.l.l.p.j.ClockBenchmark.testBaseline
sample 238251 27.986 0.132 ns/op
o.a.l.l.p.j.ClockBenchmark.testCachedClockSimpleMessage
sample 265004 546.528 2.022 ns/op
o.a.l.l.p.j.ClockBenchmark.testCachedClockSimpleMessageNoDowncast
sample 265057 192.856 0.531 ns/op
o.a.l.l.p.j.ClockBenchmark.testCachedClockTimestampMessage
sample 278932 65.167 0.191 ns/op
o.a.l.l.p.j.ClockBenchmark.testCoarseCachedClockSimpleMessage
sample 358047 85.017 0.122 ns/op
o.a.l.l.p.j.ClockBenchmark.testCoarseCachedClockSimpleMessageNoDowncast
sample 366842 84.530 0.168 ns/op
o.a.l.l.p.j.ClockBenchmark.testCoarseCachedClockTimestampMessage
sample 255360 69.229 0.208 ns/op
o.a.l.l.p.j.ClockBenchmark.testFakeClockSimpleMessage
sample 272217 105.782 0.199 ns/op
o.a.l.l.p.j.ClockBenchmark.testFakeClockSimpleMessageNoDowncast
sample 365182 85.052 0.165 ns/op
o.a.l.l.p.j.ClockBenchmark.testFakeClockTimestampMessage
sample 286814 65.450 0.282 ns/op
o.a.l.l.p.j.ClockBenchmark.testOrdinaryLogEvent
sample 199051 136.273 0.277 ns/op
o.a.l.l.p.j.ClockBenchmark.testSystemClockSimpleMessage
sample 297985 135.690 0.312 ns/op
o.a.l.l.p.j.ClockBenchmark.testSystemClockSimpleMessageNoDowncast
sample 235139 124.363 0.176 ns/op
o.a.l.l.p.j.ClockBenchmark.testSystemClockTimestampMessage
sample 290519 64.377 0.231 ns/op
{noformat}
> Avoid unnecessary Clock calls when TimestampMessage is logged
> -------------------------------------------------------------
>
> Key: LOG4J2-744
> URL: https://issues.apache.org/jira/browse/LOG4J2-744
> Project: Log4j 2
> Issue Type: Improvement
> Reporter: Scott Harrington
> Priority: Minor
> Attachments: ClockBenchmark-jdk1.7.0_45.txt, ClockBenchmark.java,
> LOG4J2-744-test.patch, LOG4J2-744.patch
>
>
> The TimestampMessage interface was introduced in LOG4J2-53 and revised for
> AsyncLogger in LOG4J2-455.
> I've observed that Clock.currentTimeMillis is still called which should not
> be necessary.
> I have two patches, one which adds JUnit tests that demonstrate the
> unnecessary Clock calls, and one which fixes the issue for both AsyncLogger
> and "traditional" configurations.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
