Ctrl+K

Sparse Linear Algebra

Sparse Linear Algebra#

See the latest specification for the sparse domain here.

This page documents implementation specific or backend specific details of the sparse domain.

OneMKL Intel CPU and GPU backends#

Currently known limitations:

  • All operations’ algorithms except no_optimize_alg map to the default algorithm.

  • The required external workspace size is always 0 bytes.

  • oneapi::math::sparse::set_csr_data and oneapi::math::sparse::set_coo_data functions cannot be used on a handle that has already been used for an operation or its optimize function. Doing so will throw a oneapi::math::unimplemented exception.

  • Using spsv with the oneapi::math::sparse::spsv_alg::no_optimize_alg and a sparse matrix that does not have the oneapi::math::sparse::matrix_property::sorted property will throw a oneapi::math::unimplemented exception.

  • Using spmm on Intel GPU with a sparse matrix that is oneapi::math::transpose::conjtrans and has the oneapi::math::sparse::matrix_property::symmetric property will throw a oneapi::math::unimplemented exception.

  • Using spmv with a sparse matrix that is oneapi::math::transpose::conjtrans with a type_view matrix_descr::symmetric or matrix_descr::hermitian will throw a oneapi::math::unimplemented exception.

  • Using spsv on Intel GPU with a sparse matrix that is oneapi::math::transpose::conjtrans and will throw a oneapi::math::unimplemented exception.

  • Scalar parameters alpha and beta should be host pointers to prevent synchronizations and copies to the host.

cuSPARSE backend#

Currently known limitations:

  • The COO format requires the indices to be sorted by row. See the cuSPARSE documentation. Sparse operations using matrices with the COO format without the property matrix_property::sorted_by_rows or matrix_property::sorted will throw a oneapi::math::unimplemented exception.

  • Using spmm with the algorithm spmm_alg::csr_alg3 and an opA other than transpose::nontrans or an opB transpose::conjtrans will throw a oneapi::math::unimplemented exception.

  • Using spmm with the algorithm spmm_alg::csr_alg3, opB=transpose::trans and real fp64 precision will throw a oneapi::math::unimplemented exception. This configuration can fail as of CUDA 12.6.2, see the related issue `here<https://forums.developer.nvidia.com/t/cusparse-spmm-sample-failing-with-misaligned-address/311022>`_.

  • Using spmv with a type_view other than matrix_descr::general will throw a oneapi::math::unimplemented exception.

  • Using spsv with the algorithm spsv_alg::no_optimize_alg may still perform some mandatory preprocessing.

  • oneMath does not provide a way to use non-default algorithms without calling preprocess functions such as cusparseSpMM_preprocess or cusparseSpMV_preprocess. Feel free to create an issue if this is needed.

rocSPARSE backend#

Currently known limitations:

  • Using spmv with a type_view other than matrix_descr::general will throw a oneapi::math::unimplemented exception.

  • The COO format requires the indices to be sorted by row then by column. See the rocSPARSE COO documentation. Sparse operations using matrices with the COO format without the property matrix_property::sorted will throw a oneapi::math::unimplemented exception.

  • The CSR format requires the column indices to be sorted within each row. See the rocSPARSE CSR documentation. Sparse operations using matrices with the CSR format without the property matrix_property::sorted will throw a oneapi::math::unimplemented exception.

  • The same sparse matrix handle cannot be reused for multiple operations spmm, spmv, or spsv. Doing so will throw a oneapi::math::unimplemented exception. See #332.

Operation algorithms mapping#

The following tables describe how a oneMath algorithm maps to the backend’s algorithms. Refer to the backend’s documentation for a more detailed explanation of the algorithms.

Backends with no equivalent algorithms will fallback to the backend’s default behavior.

spmm#

spmv#

spsv#

Contents