If an arm core doesn't support VFP, so it cannot support float number compute?
Correct. If the processor does not have hardware support for floating point computation, then this would have to be supported via software libraries.
Chris