On Thu, 06 Jun 2019 10:40:47 -0000 "Amitoz AZAD" <rteja...@gmail.com> wrote:
> Merci Jerome, > > So my understanding is that you extracted the row , column indices and the > data from the sparse matrix and then passed it as the normal numpy vectors? > lut (3 tuple) the variable is your sparse matrix ? Hi Amitoz, Yes, this "LUT" is a 3-tuple containing the CSR representation of the sparse matrix. You can use scipy.sparse.csr_matrix to generate the sparse represenation and retrieve those "data", "indices" and "indptr" vectors which are compatible with my code. Another way to do this is to use a sparsifyer on the the device, but this requires the full matrix fits into the device's memory. https://github.com/silx-kit/silx/pull/2625 This is ongoing work ! The most usual way is neverthless to build those vector directly when the sparse matrix is too large to fit into memory. Cheers, Jerome _______________________________________________ PyOpenCL mailing list -- pyopencl@tiker.net To unsubscribe send an email to pyopencl-le...@tiker.net
