same thing (as the VC4 seems not to have double-precision float support). --HG-- branch : dtrg-videocore