Synchronization article on developerWorks

Bryce McKinlay
Tue Jul 24 16:14:00 GMT 2001

On Wednesday, July 25, 2001, at 10:47  AM, Jeff Sturm wrote:

> I don't understand how this author got the results he did.  It doesn't
> look like he made his source available.

Trying to measure the cost of synchronization by timing a synchronized 
vs un-synchronized call means very little anyhow. A better test would be 
to measure the performance of some real-world single-threaded 
application with synchronization enabled vs the same application with 
syncronization removed. Of course, turning off synchronization is not 
easy with most JVMs.



More information about the Java mailing list