I am trying to calculate a FFT with pyFFT. But the following code results in the error.
self.ctx = cl.create_some_context(interactive=False)
self.queue = cl.CommandQueue(self.ctx)
self.plan = Plan(fft_size, self.queue)
gpu_data0 = cl_array.to_device(self.ctx, self.queue, ref_input)
ref_input is of type:
< type 'numpy.ndarray' >
debug output:
< pyopencl._cl.CommandQueue object at 0x7f0686c10628 >
['__class__', '__delattr__', '__dict__', '__doc__', '__enter__', '__eq__',
'__exit__','__format__', '__getattribute__', '__hash__', '__init__',
'__instance_size__', '__module__', '__ne__', '__new__', '__reduce__',
'__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__',
'__subclasshook__', '__weakref__', '_get_cl_version', 'context', 'device',
'finish', 'flush', 'from_int_ptr', 'get_info', 'int_ptr', 'properties',
'reference_count']
Traceback (most recent call last):
File "../dsp.py", line 200, in gpu_fft
gpu_data0 = cl_array.to_device(self.ctx, self.queue, ref_data)
File "/usr/lib/python2.7/dist-packages/pyopencl/array.py", line 1446, in to_device
if ary.dtype == object:
AttributeError: 'CommandQueue' object has no attribute 'dtype'
In fact I cannot find 'dtype'. But why?
The function pyopencl.array.to_device does not take the context as a parameter according to the documentation. Use this instead:
gpu_data0 = cl_array.to_device(self.queue, ref_input)