Function cusparseSbsrmv 

pub unsafe extern "C" fn cusparseSbsrmv(
    handle: cusparseHandle_t,
    dirA: cusparseDirection_t,
    transA: cusparseOperation_t,
    mb: c_int,
    nb: c_int,
    nnzb: c_int,
    alpha: *const f32,
    descrA: cusparseMatDescr_t,
    bsrSortedValA: *const f32,
    bsrSortedRowPtrA: *const c_int,
    bsrSortedColIndA: *const c_int,
    blockDim: c_int,
    x: *const f32,
    beta: *const f32,
    y: *mut f32,
) -> cusparseStatus_t

This function performs the matrix-vector operation

where $A\text{ is an }(mb \ast blockDim) \times (nb \ast blockDim)$ sparse matrix that is defined in BSR storage format by the three arrays bsrVal, bsrRowPtr, and bsrColInd); x and y are vectors; $\alpha\text{ and }\beta$ are scalars; and: $$ \operatorname{op}(A) = \begin{cases} A & \text{if } trans = \text{CUSPARSE_OPERATION_NON_TRANSPOSE} \ A^T & \text{if } trans = \text{CUSPARSE_OPERATION_TRANSPOSE} \ A^H & \text{if } trans = \text{CUSPARSE_OPERATION_CONJUGATE_TRANSPOSE} \end{cases} $$

bsrmv() has the following properties:

• The routine requires no extra storage.
• The routine supports asynchronous execution.
• The routine supports CUDA graph capture.

Several comments on bsrmv():

• Only blockDim > 1 is supported

• Only cusparseOperation_t::CUSPARSE_OPERATION_NON_TRANSPOSE is supported, that is

• Only cusparseMatrixType_t::CUSPARSE_MATRIX_TYPE_GENERAL is supported.

• The size of vector x should be $(nb \ast blockDim)$ at least, and the size of vector y should be $(mb \ast blockDim)$ at least; otherwise, the kernel may return cusparseStatus_t::CUSPARSE_STATUS_EXECUTION_FAILED because of an out-of-bounds array.

For example, suppose the user has a CSR format and wants to try bsrmv(), the following code demonstrates how to use csr2bsr() conversion and bsrmv() multiplication in single precision.