Created by: wojtuss

In python/paddle/fluid/parallel_executor.py, in the ParallelExecutor's constructor there is an incrorrect call to CPUPlace's constructor from paddle/fluid/platform/place.h. CPUPlace() accepts no arguments. In ParallelExecutor it is called with an int argument:

        ...
        if use_cuda:
            for i in xrange(core.get_cuda_device_count()):
                p = core.Place()
                self._act_places.append(core.CUDAPlace(i))
                p.set_place(self._act_places[-1])
                self._places.append(p)
        else:
            for i in xrange(multiprocessing.cpu_count()):
                p = core.Place()
                self._act_places.append(core.CPUPlace(i))
                p.set_place(self._act_places[-1])
                self._places.append(p)
        assert self._places, "no place for execution"
        ...