Open whatever1983 opened 6 years ago
I would imagine it would be a lot easier to re-implement the algorithms in a GPU friendly manner, since this is currently just a lightweight wrapper around the C library.
hi, beside np.array, it can be used to do computations on something like torch.tensor (pytorch), nd.array(mxnet) and so on which are deep learning libraries and support GPU.
Just a question: since this is a Python wrapper, is there a way to accelerate this python wrapper with Numba and run it on Nvidia GPUs? Or is it a better idea to simply port the Ta-Lib C files to CUDA and use nvcc compiler instead