How to enable Hyper-Q on Tesla K20

I want to my legacy MPI program with Hyper-Q.

To begin with, I have tried to run the simple Hyper-Q example shown below.
http://docs.nvidia.com/cuda/samples/6_Advanced/simpleHyperQ/doc/HyperQ.pdf

In the above document, Visual Profiler shows that 32 streams run cuncurrently.
But in my enviroment, only 8 streams are executed at the same time.

According to this article,
http://blogs.nvidia.com/2012/08/unleash-legacy-mpi-codes-with-keplers-hyper-q/
some enviroment variable settings seems to be needed.

May I have any idea to enable Hyper-Q?
How to set the enviroment variable, etc.

Regards