np_float16x2

nvmath.device.np_float16x2 = dtype([('x', '<f2'), ('y', '<f2')], align=True)

NumPy dtype object that encapsulates the np in device.

Parameters:

• x

• y