[GitHub] [logging-log4j2] vy edited a comment on issue #335: Import of LogstashLayout as JsonTemplateLayout

[GitBox <gi...@apache.org>]

vy edited a comment on issue #335: Import of LogstashLayout as JsonTemplateLayout
URL: https://github.com/apache/logging-log4j2/pull/335#issuecomment-580189684
 
 
   > The NoOp benchmark should be renamed Allocate
   
   Check.
   
   > I don't believe that `MpmcArrayQueue` is lock-free
   
   The definition of what is lock-free is quite [a debate](https://groups.google.com/forum/#!topic/lock-free/MnhH9AlLfOc) on its own. But according to [README](/JCTools/JCTools#jctools), `MPMC` is lock-less (what ever that means) and bounded.
   
   I am just baffled by the fact how come `AllocatePool` can yield such a miraculous performance that is almost on par with `ThreadLocalPool`. I have shared this question with @richardstartin and he kindly gave the following feedback:
   
   > I think that the G1 write barrier is the cost in the MPMC case (I haven't run the benchmark though) and I would be interested to know the marginal cost of G1 in this case. The most interesting metric is MPMC/G1 vs MPMC/other GC
   
   I will experiment with smaller heap sizes, bigger `StringBuilder` capacities, and different GCs.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services