concurent access to HashMap

[Юрий Михеев]

I try reading and writing to HTreeMap from 5 threads And have 75% CPU time locked is it normal? How can I resolve it?

[Jan Kotek]

This has been fixed in 3.0.5, there was performance issue

http://www.mapdb.org/changelog/

--
You received this message because you are subscribed to the Google Groups "MapDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mapdb+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Юрий Михеев]

Jan, thank for you work and the answer, I checked the version I have. It is 3.1.0-SHAPSHOT

[Jan Kotek]

In that case switch to older version, snapshot may not have this fix

[Юрий Михеев]

I did it. Now 72% CPU time is spent on org.mapdb.HTreeMap.get

in a case of only one thread org.mapdb.HTreeMap.get used 2,3% CPU time.

in fact, I think the lock  is in version 3.0.5 too

[Юрий Михеев]

I realized the reason is my data structure. I have a small set of keys which occur too often. So getting and setting values blocks each other.