CuPy Example

This example can be found, along with several others, at the CuPy github site.

Discussion

CuPy implements many NumPy built-in functions on the GPU using a very similar syntax. It has a few distinct features such as device specification and data transfer between host and device.

This example generates a random matrix A and vector x, uses them to compute a vector b, then solves the linear system Ax=b using the computed b as the right-hand side, then compares the computed b to the constructed one. It is mostly like NumPy and the fit function can be used with either CPU-based arrays or GPU-based arrays. The run routine uses np.astype to generate the ndarrays on the host and cupy.asarray to generate equivalents on the GPU; asarray allocated device memory. The get_array_module function returns whether the array argument is a NumPy or CuPy ndarray, and the new vector is assigned to the same location.