Math operations optimization

Hi.

I'm creating an application in which I create in the device a RSA certificate with bouncy castle.

Everything works as expected, but generating a 4096b key takes 10-15 minutes in a normal device.

I know generating a RSA key is a very cpu intensive process, but I was wondering if there is anything what can be passed to the compiler so it optimizes those math operations.

Any idea?

Thanks in advance