PerformanceTestValidator purpose

Dear All,

when building Batik from sources, regardless ant or maven, it fails on the
following performance test:

batik-test/src/main/java/org/apache/batik/test/PerformanceTestValidator.java

"unexpected.performance.test.failure"

There are several runs (p.run()) of the same simple test. If there are huge
differences in the duration, the test fails. 

On my computer there are realy big differences in even and odd runs, but
both even and odd are roughly similar. I can't see any reason for this
behavior.

But I also fail to understand the purpose of this test. Why we evaluate
improvements / regression the same code running in the loop?

Thanks for any clarification,

Jan