"Terry Reedy" <[EMAIL PROTECTED]> wrote:

> "Tom Anderson" <[EMAIL PROTECTED]> wrote:
> > I can't immediately see any properties of this network that could be
> > exploited, but that doesn't mean there aren't any.
> No it doesn't.  The challenge is to find a property that saves more time,
> across trials, that it takes to compute.

There could have been one if the schedule was fixed across trials. However, if 
you look into the
test script, this is not the case; a new schedule is randomly generated every 
time. So any heavy
preprocessing on the schedule is rather unlikely to pay off.



Reply via email to