Github user dkuppitz commented on the issue:
https://github.com/apache/tinkerpop/pull/948
That's the kind of benchmark I was looking for. I don't think it needs to
be part of the project (these micro-benchmarks always seem to be too unstable
and making them part of the test suite usually just makes the builds fail at
some point) - thus a manual test is good enough.
VOTE + 1---
