I have access to 30-day Trial Editions of the ARM RealView toolchain
which has a great instruction set simulator.
It's free to BeagleBoard hobbyists, but you'll need to pay for
shipping. Note that I have to pass on your name and contact
information to ARM as part of my agreement with them which allows me
to do this.
The ARM ISS must support ARMv7-A as part of the NEON instuction set
support. You're welcome to get a copy to try out for yourself to see
how good that support really is.
I'm putting together a virtual/hybrid simulation environment of the
BeagleBoard to help with various code optimization experiments.
Perhaps you can help in this effort.