PyOpenCL has two ways to create an OpenCL buffer:
pyopencl.Buffer takes a numpy array and turns it into a buffer.
pyopencl.array.Array takes a numpy array and turns it into a PyOpenCL array - an object that can still be edited as a numpy array but provides a buffer when you call .data on it.
Is there a performance / function reason for choosing one or the other? Or should I just choose the one that makes my code more readable?
source share