Unfortunately we had done a mistake in AOP benchmark article we published a couple of days ago. The mistake was when implementing some of the interceptors for dynaop (we used the API the wrong way when we wanted to access the proxied, 'target instance').

Before we accessed the 'target instance' like this:

    Object target = invocation.getProxy().getProxyContext().unwrap();

    Object target = invocation.getProxy();

When correcting these mistakes dynaop performs up to 10 times better in the situations where we are accessing the 'target instance'.

The article has been updated with the new figures.

Thanks to Bob Lee who helped us discover this flaw.

/The AspectWerkz Team