Yet another strike against the current AA implementation

[dsimcha]

== Quote from Robert Fraser (fraserofthenight at gmail.com)'s article
> Did you compile that with optimization/inlining? If not, it's not really
> a fair test... and if so why isn't the compiler getting rid of those
> pointless loops?

Yes, -O -inline -release.  I didn't bother stating that explicitly b/c it's the
standard way of doing benchmarks.