There are fixed-point models too which can be deterministic. But for floating point you are very instruction-set dependent, never mind floating point operations

There are fixed-point models too which can be deterministic. But for floating point you are very instruction-set dependent, never mind floating point operations not being commutative.