This is the PerfTestDriver test class (in log4j-core/test, package ...async.perf). Mainly perf3PlainNoLocation.xml: RollingRandomAccessFileAppender, PatternLayout, all loggers are AsyncLoggers, logging a simple string without parameters.
Profiling with YourKit did not tell me anything useful. I'm now eliminating the effect of Layouts/Appenders, using CountingNoOpAppender, and seeing similar numbers. So this seems to be mostly an issue in AsyncLogger. I'll let you know when I find out more. There's a lot of trial and error here, so this may take a while... Remko Sent from my iPhone > On 2016/02/26, at 21:02, Mikael Ståldal <mikael.stal...@magine.com> wrote: > > Which components (appenders, layouts) are involved in the tests? Would it be > possible to do some profiling to see if there is any particular component > which is to blame? > >> On Fri, Feb 26, 2016 at 12:51 PM, Remko Popma <remko.po...@gmail.com> wrote: >> To give you some rough impression on concrete numbers for this trend: >> 2.0: ~6M ops/sec >> 2.1-2.2: ~5M ops/sec >> 2.3-2.4: ~3-4M ops/sec >> 2.5: ~3M ops/sec >> 2.6: ~2M ops/sec >> >> >>> On Friday, 26 February 2016, Remko Popma <remko.po...@gmail.com> wrote: >>> You're absolutely right. I still have quite a few unit tests to add. >>> >>> Initial perf testing shows a downward trend in Async Logger performance >>> with every release. (Logging simple string messages without params.) This >>> is worrisome and I'm focusing on figuring that out first: this will likely >>> involve additional code changes and I'll add more tests after that. >>> >>> Sent from my iPhone >>> >>>> On 2016/02/26, at 10:38, Gary Gregory <garydgreg...@gmail.com> wrote: >>>> >>>> Wow, I love the activity we are seeing toward 2.6! All the perf work on >>>> top of an existing sizable change set. Very exciting indeed. >>>> >>>> There sure are a lot of changes coming in. I hope that we all can pitch in >>>> to make sure most if not all of these changes get code coverage from unit >>>> tests. I've not checked closely, but it seems like we may not have good >>>> coverage _yet_, or do I have the wrong impression? >>>> >>>> I want to make sure we keep our stability in tip top shape :-) and that we >>>> have no regression from previous releases. >>>> >>>> Gary >>>> >>>> -- >>>> E-Mail: garydgreg...@gmail.com | ggreg...@apache.org >>>> Java Persistence with Hibernate, Second Edition >>>> JUnit in Action, Second Edition >>>> Spring Batch in Action >>>> Blog: http://garygregory.wordpress.com >>>> Home: http://garygregory.com/ >>>> Tweet! http://twitter.com/GaryGregory > > > > -- > > > Mikael Ståldal > Senior software developer > > Magine TV > mikael.stal...@magine.com > Grev Turegatan 3 | 114 46 Stockholm, Sweden | www.magine.com > > Privileged and/or Confidential Information may be contained in this message. > If you are not the addressee indicated in this message > (or responsible for delivery of the message to such a person), you may not > copy or deliver this message to anyone. In such case, > you should destroy this message and kindly notify the sender by reply email. >
