Sparse Matrix APIs

The cuSPARSE helper functions for sparse matrix descriptor are described in this section.

See the COO, CSR , CSC , SELL , BSR, Blocked-Ell sections for the detailed description of the storage formats.

Coordinate (COO)

cusparseCreateCoo()

cusparseStatus_t
cusparseCreateCoo(cusparseSpMatDescr_t* spMatDescr,
                  int64_t               rows,
                  int64_t               cols,
                  int64_t               nnz,
                  void*                 cooRowInd,
                  void*                 cooColInd,
                  void*                 cooValues,
                  cusparseIndexType_t   cooIdxType,
                  cusparseIndexBase_t   idxBase,
                  cudaDataType          valueType)

cusparseStatus_t
cusparseCreateConstCoo(cusparseConstSpMatDescr_t* spMatDescr,
                       int64_t                    rows,
                       int64_t                    cols,
                       int64_t                    nnz,
                       const void*                cooRowInd,
                       const void*                cooColInd,
                       const void*                cooValues,
                       cusparseIndexType_t        cooIdxType,
                       cusparseIndexBase_t        idxBase,
                       cudaDataType               valueType)

This function initializes the sparse matrix descriptor spMatDescr in the COO format (Structure of Arrays layout).

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	OUT	Sparse matrix descriptor
`rows`	HOST	IN	Number of rows of the sparse matrix
`cols`	HOST	IN	Number of columns of the sparse matrix
`nnz`	HOST	IN	Number of non-zero entries of the sparse matrix
`cooRowInd`	DEVICE	IN	Row indices of the sparse matrix. Array with `nnz` elements
`cooColInd`	DEVICE	IN	Column indices of the sparse matrix. Array with `nnz` elements
`cooValues`	DEVICE	IN	Values of the sparse matrix. Array with `nnz` elements
`cooIdxType`	HOST	IN	Data type of `cooRowInd` and `cooColInd`
`idxBase`	HOST	IN	Index base of `cooRowInd` and `cooColInd`
`valueType`	HOST	IN	Datatype of `cooValues`

cusparseCreateCoo() has the following constraints:

cooRowInd, cooColInd, and cooValues must be aligned to the size of the datatypes specified by cooIdxType, cooIdxType, and valueType. respectively. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

cusparseCooGet()

cusparseStatus_t
cusparseCooGet(cusparseSpMatDescr_t spMatDescr,
               int64_t*             rows,
               int64_t*             cols,
               int64_t*             nnz,
               void**               cooRowInd,
               void**               cooColInd,
               void**               cooValues,
               cusparseIndexType_t* idxType,
               cusparseIndexBase_t* idxBase,
               cudaDataType*        valueType)

cusparseStatus_t
cusparseConstCooGet(cusparseConstSpMatDescr_t spMatDescr,
                    int64_t*                  rows,
                    int64_t*                  cols,
                    int64_t*                  nnz,
                    const void**              cooRowInd,
                    const void**              cooColInd,
                    const void**              cooValues,
                    cusparseIndexType_t*      idxType,
                    cusparseIndexBase_t*      idxBase,
                    cudaDataType*             valueType)

This function returns the fields of the sparse matrix descriptor spMatDescr stored in COO format (Array of Structures layout).

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`rows`	HOST	OUT	Number of rows of the sparse matrix
`cols`	HOST	OUT	Number of columns of the sparse matrix
`nnz`	HOST	OUT	Number of non-zero entries of the sparse matrix
`cooRowInd`	DEVICE	OUT	Row indices of the sparse matrix. Array `nnz` elements
`cooColInd`	DEVICE	OUT	Column indices of the sparse matrix. Array `nnz` elements
`cooValues`	DEVICE	OUT	Values of the sparse matrix. Array `nnz` elements
`cooIdxType`	HOST	OUT	Data type of `cooRowInd` and `cooColInd`
`idxBase`	HOST	OUT	Index base of `cooRowInd` and `cooColInd`
`valueType`	HOST	OUT	Datatype of `cooValues`

See cusparseStatus_t for the description of the return status.

cusparseCooSetPointers()

cusparseStatus_t
cusparseCooSetPointers(cusparseSpMatDescr_t spMatDescr,
                       void*                cooRows,
                       void*                cooColumns,
                       void*                cooValues)

This function sets the pointers of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`cooRows`	DEVICE	IN	Row indices of the sparse matrix. Array with `nnz` elements
`cooColumns`	DEVICE	IN	Column indices of the sparse matrix. Array with `nnz` elements
`cooValues`	DEVICE	IN	Values of the sparse matrix. Array with `nnz` elements

cusparseCooSetPointers() has the following constraints:

cooRows, cooColumns, and cooValues must be aligned to the size of their corresponding datatypes specified in spMatDescr. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

cusparseCooSetStridedBatch()

cusparseStatus_t
cusparseCooSetStridedBatch(cusparseSpMatDescr_t spMatDescr,
                           int                  batchCount,
                           int64_t              batchStride)

This function sets the batchCount and the batchStride fields of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`batchCount`	HOST	IN	Number of batches of the sparse matrix
`batchStride`	HOST	IN	address offset between consecutive batches

See cusparseStatus_t for the description of the return status.

Compressed Sparse Row (CSR)

cusparseCreateCsr()

cusparseStatus_t
cusparseCreateCsr(cusparseSpMatDescr_t* spMatDescr,
                  int64_t               rows,
                  int64_t               cols,
                  int64_t               nnz,
                  void*                 csrRowOffsets,
                  void*                 csrColInd,
                  void*                 csrValues,
                  cusparseIndexType_t   csrRowOffsetsType,
                  cusparseIndexType_t   csrColIndType,
                  cusparseIndexBase_t   idxBase,
                  cudaDataType          valueType)

cusparseStatus_t
cusparseCreateConstCsr(cusparseConstSpMatDescr_t* spMatDescr,
                       int64_t                    rows,
                       int64_t                    cols,
                       int64_t                    nnz,
                       const void*                csrRowOffsets,
                       const void*                csrColInd,
                       const void*                csrValues,
                       cusparseIndexType_t        csrRowOffsetsType,
                       cusparseIndexType_t        csrColIndType,
                       cusparseIndexBase_t        idxBase,
                       cudaDataType               valueType)

This function initializes the sparse matrix descriptor spMatDescr in the CSR format.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	OUT	Sparse matrix descriptor
`rows`	HOST	IN	Number of rows of the sparse matrix
`cols`	HOST	IN	Number of columns of the sparse matrix
`nnz`	HOST	IN	Number of non-zero entries of the sparse matrix
`csrRowOffsets`	DEVICE	IN	Row offsets of the sparse matrix. Array with `rows + 1` elements
`csrColInd`	DEVICE	IN	Column indices of the sparse matrix. Array with `nnz` elements
`csrValues`	DEVICE	IN	Values of the sparse matrix. Array with `nnz` elements
`csrRowOffsetsType`	HOST	IN	Data type of `csrRowOffsets`
`csrColIndType`	HOST	IN	Data type of `csrColInd`
`idxBase`	HOST	IN	Index base of `csrRowOffsets` and `csrColInd`
`valueType`	HOST	IN	Datatype of `csrValues`

cusparseCreateCsr() has the following constraints:

csrRowOffsets, csrColInd, and csrValues must be aligned to the size of the datatypes specified by csrRowOffsetsType, csrColIndType, and valueType, respectively. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

cusparseCsrGet()

cusparseStatus_t
cusparseCsrGet(cusparseSpMatDescr_t spMatDescr,
               int64_t*             rows,
               int64_t*             cols,
               int64_t*             nnz,
               void**               csrRowOffsets,
               void**               csrColInd,
               void**               csrValues,
               cusparseIndexType_t* csrRowOffsetsType,
               cusparseIndexType_t* csrColIndType,
               cusparseIndexBase_t* idxBase,
               cudaDataType*        valueType)

cusparseStatus_t
cusparseConstCsrGet(cusparseConstSpMatDescr_t spMatDescr,
                    int64_t*                  rows,
                    int64_t*                  cols,
                    int64_t*                  nnz,
                    const void**              csrRowOffsets,
                    const void**              csrColInd,
                    const void**              csrValues,
                    cusparseIndexType_t*      csrRowOffsetsType,
                    cusparseIndexType_t*      csrColIndType,
                    cusparseIndexBase_t*      idxBase,
                    cudaDataType*             valueType)

This function returns the fields of the sparse matrix descriptor spMatDescr stored in CSR format.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`rows`	HOST	OUT	Number of rows of the sparse matrix
`cols`	HOST	OUT	Number of columns of the sparse matrix
`nnz`	HOST	OUT	Number of non-zero entries of the sparse matrix
`csrRowOffsets`	DEVICE	OUT	Row offsets of the sparse matrix. Array with `rows + 1` elements
`csrColInd`	DEVICE	OUT	Column indices of the sparse matrix. Array with `nnz` elements
`csrValues`	DEVICE	OUT	Values of the sparse matrix. Array with `nnz` elements
`csrRowOffsetsType`	HOST	OUT	Data type of `csrRowOffsets`
`csrColIndType`	HOST	OUT	Data type of `csrColInd`
`idxBase`	HOST	OUT	Index base of `csrRowOffsets` and `csrColInd`
`valueType`	HOST	OUT	Datatype of `csrValues`

See cusparseStatus_t for the description of the return status.

cusparseCsrSetPointers()

cusparseStatus_t
cusparseCsrSetPointers(cusparseSpMatDescr_t spMatDescr,
                       void*                csrRowOffsets,
                       void*                csrColInd,
                       void*                csrValues)

This function sets the pointers of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`csrRowOffsets`	DEVICE	IN	Row offsets of the sparse matrix. Array with `rows + 1` elements
`csrColInd`	DEVICE	IN	Column indices of the sparse matrix. Array with `nnz` elements
`csrValues`	DEVICE	IN	Values of the sparse matrix. Array with `nnz` elements

cusparseCsrSetPointers() has the following constraints:

csrRowOffsets, csrColInd, and csrValues must be aligned to the size of their corresponding datatypes specified in spMatDescr. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

cusparseCsrSetStridedBatch()

cusparseStatus_t
cusparseCsrSetStridedBatch(cusparseSpMatDescr_t spMatDescr,
                           int                  batchCount,
                           int64_t              offsetsBatchStride,
                           int64_t              columnsValuesBatchStride)

This function sets the batchCount and the batchStride fields of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`batchCount`	HOST	IN	Number of batches of the sparse matrix
`offsetsBatchStride`	HOST	IN	Address offset between consecutive batches for the row offset array
`columnsValuesBatchStride`	HOST	IN	Address offset between consecutive batches for the column and value arrays

See cusparseStatus_t for the description of the return status.

Compressed Sparse Column (CSC)

cusparseCreateCsc()

cusparseStatus_t
cusparseCreateCsc(cusparseSpMatDescr_t* spMatDescr,
                  int64_t               rows,
                  int64_t               cols,
                  int64_t               nnz,
                  void*                 cscColOffsets,
                  void*                 cscRowInd,
                  void*                 cscValues,
                  cusparseIndexType_t   cscColOffsetsType,
                  cusparseIndexType_t   cscRowIndType,
                  cusparseIndexBase_t   idxBase,
                  cudaDataType          valueType)

cusparseStatus_t
cusparseCreateConstCsc(cusparseConstSpMatDescr_t* spMatDescr,
                       int64_t                    rows,
                       int64_t                    cols,
                       int64_t                    nnz,
                       const void*                cscColOffsets,
                       const void*                cscRowInd,
                       const void*                cscValues,
                       cusparseIndexType_t        cscColOffsetsType,
                       cusparseIndexType_t        cscRowIndType,
                       cusparseIndexBase_t        idxBase,
                       cudaDataType               valueType)

This function initializes the sparse matrix descriptor spMatDescr in the CSC format.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	OUT	Sparse matrix descriptor
`rows`	HOST	IN	Number of rows of the sparse matrix
`cols`	HOST	IN	Number of columns of the sparse matrix
`nnz`	HOST	IN	Number of non-zero entries of the sparse matrix
`cscColOffsets`	DEVICE	IN	Column offsets of the sparse matrix. Array with `cols + 1` elements
`cscRowInd`	DEVICE	IN	Row indices of the sparse matrix. Array with `nnz` elements
`cscValues`	DEVICE	IN	Values of the sparse matrix. Array with `nnz` elements
`cscColOffsetsType`	HOST	IN	Data type of `cscColOffsets`
`cscRowIndType`	HOST	IN	Data type of `cscRowInd`
`idxBase`	HOST	IN	Index base of `cscColOffsets` and `cscRowInd`
`valueType`	HOST	IN	Datatype of `cscValues`

cusparseCreateCsc() has the following constraints:

cscColOffsets, cscRowInd, and cscValues must be aligned to the size of the datatypes specified by cscColOffsetsType, cscRowIndType, and valueType, respectively. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

cusparseCscGet()

cusparseStatus_t
cusparseCscGet(cusparseSpMatDescr_t spMatDescr,
               int64_t*             rows,
               int64_t*             cols,
               int64_t*             nnz,
               void**               cscColOffsets,
               void**               cscRowInd,
               void**               cscValues,
               cusparseIndexType_t* cscColOffsetsType,
               cusparseIndexType_t* cscRowIndType,
               cusparseIndexBase_t* idxBase,
               cudaDataType*        valueType)

cusparseStatus_t
cusparseConstCscGet(cusparseConstSpMatDescr_t spMatDescr,
                    int64_t*                  rows,
                    int64_t*                  cols,
                    int64_t*                  nnz,
                    const void**              cscColOffsets,
                    const void**              cscRowInd,
                    const void**              cscValues,
                    cusparseIndexType_t*      cscColOffsetsType,
                    cusparseIndexType_t*      cscRowIndType,
                    cusparseIndexBase_t*      idxBase,
                    cudaDataType*             valueType)

This function returns the fields of the sparse matrix descriptor spMatDescr stored in CSC format.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`rows`	HOST	OUT	Number of rows of the sparse matrix
`cols`	HOST	OUT	Number of columns of the sparse matrix
`nnz`	HOST	OUT	Number of non-zero entries of the sparse matrix
`cscColOffsets`	DEVICE	OUT	Row offsets of the sparse matrix. Array with `cols + 1` elements
`cscRowInd`	DEVICE	OUT	Column indices of the sparse matrix. Array with `nnz` elements
`cscValues`	DEVICE	OUT	Values of the sparse matrix. Array with `nnz` elements
`cscColOffsetsType`	HOST	OUT	Data type of `cscColOffsets`
`cscRowIndType`	HOST	OUT	Data type of `cscRowInd`
`idxBase`	HOST	OUT	Index base of `cscColOffsets` and `cscRowInd`
`valueType`	HOST	OUT	Datatype of `cscValues`

See cusparseStatus_t for the description of the return status.

cusparseCscSetPointers()

cusparseStatus_t
cusparseCscSetPointers(cusparseSpMatDescr_t spMatDescr,
                       void*                cscColOffsets,
                       void*                cscRowInd,
                       void*                cscValues)

This function sets the pointers of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`cscColOffsets`	DEVICE	IN	Col offsets of the sparse matrix. Array with `cols + 1` elements
`cscRowInd`	DEVICE	IN	Row indices of the sparse matrix. Array with `nnz` elements
`cscValues`	DEVICE	IN	Values of the sparse matrix. Array with `nnz` elements

cusparseCscSetPointers() has the following constraints:

cscColOffsets, cscRowInd, and cscValues must be aligned to the size of their corresponding datatypes specified in spMatDescr. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

Blocked-Ellpack (Blocked-ELL)

cusparseCreateBlockedEll()

cusparseStatus_t
cusparseCreateBlockedEll(cusparseSpMatDescr_t* spMatDescr,
                         int64_t               rows,
                         int64_t               cols,
                         int64_t               ellBlockSize,
                         int64_t               ellCols,
                         void*                 ellColInd,
                         void*                 ellValue,
                         cusparseIndexType_t   ellIdxType,
                         cusparseIndexBase_t   idxBase,
                         cudaDataType          valueType)

cusparseStatus_t
cusparseCreateConstBlockedEll(cusparseConstSpMatDescr_t* spMatDescr,
                              int64_t                    rows,
                              int64_t                    cols,
                              int64_t                    ellBlockSize,
                              int64_t                    ellCols,
                              const void*                ellColInd,
                              const void*                ellValue,
                              cusparseIndexType_t        ellIdxType,
                              cusparseIndexBase_t        idxBase,
                              cudaDataType               valueType)

This function initializes the sparse matrix descriptor spMatDescr for the Blocked-Ellpack (ELL) format.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	OUT	Sparse matrix descriptor
`rows`	HOST	IN	Number of rows of the sparse matrix
`cols`	HOST	IN	Number of columns of the sparse matrix
`ellBlockSize`	HOST	IN	Size of the ELL-Block
`ellCols`	HOST	IN	Actual number of columns of the Blocked-Ellpack format (`ellValue` columns)
`ellColInd`	DEVICE	IN	Blocked-ELL Column indices. Array with `[ellCols / ellBlockSize][rows / ellBlockSize]` elements
`ellValue`	DEVICE	IN	Values of the sparse matrix. Array with `rows * ellCols` elements
`ellIdxType`	HOST	IN	Data type of `ellColInd`
`idxBase`	HOST	IN	Index base of `ellColInd`
`valueType`	HOST	IN	Data type of `ellValue`

Blocked-ELL Column indices (ellColInd) are in the range [0, cols / ellBlockSize -1]. The array can contain -1 values for indicating empty blocks.

See cusparseStatus_t for the description of the return status.

cusparseBlockedEllGet()

cusparseStatus_t
cusparseBlockedEllGet(cusparseSpMatDescr_t spMatDescr,
                      int64_t*             rows,
                      int64_t*             cols,
                      int64_t*             ellBlockSize,
                      int64_t*             ellCols,
                      void**               ellColInd,
                      void**               ellValue,
                      cusparseIndexType_t* ellIdxType,
                      cusparseIndexBase_t* idxBase,
                      cudaDataType*        valueType)

cusparseStatus_t
cusparseConstBlockedEllGet(cusparseConstSpMatDescr_t spMatDescr,
                           int64_t*                  rows,
                           int64_t*                  cols,
                           int64_t*                  ellBlockSize,
                           int64_t*                  ellCols,
                           const void**              ellColInd,
                           const void**              ellValue,
                           cusparseIndexType_t*      ellIdxType,
                           cusparseIndexBase_t*      idxBase,
                           cudaDataType*             valueType)

This function returns the fields of the sparse matrix descriptor spMatDescr stored in Blocked-Ellpack (ELL) format.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`rows`	HOST	OUT	Number of rows of the sparse matrix
`cols`	HOST	OUT	Number of columns of the sparse matrix
`ellBlockSize`	HOST	OUT	Size of the ELL-Block
`ellCols`	HOST	OUT	Actual number of columns of the Blocked-Ellpack format
`ellColInd`	DEVICE	OUT	Column indices for the ELL-Block. Array with `[cols / ellBlockSize][rows / ellBlockSize]` elements
`ellValue`	DEVICE	OUT	Values of the sparse matrix. Array with `rows * ellCols` elements
`ellIdxType`	HOST	OUT	Data type of `ellColInd`
`idxBase`	HOST	OUT	Index base of `ellColInd`
`valueType`	HOST	OUT	Datatype of `ellValue`

See cusparseStatus_t for the description of the return status.

Sliced-Ellpack (SELL)

cusparseCreateSlicedEll()

cusparseStatus_t
cusparseCreateSlicedEll(cusparseSpMatDescr_t*   spMatDescr,
                        int64_t                 rows,
                        int64_t                 cols,
                        int64_t                 nnz,
                        int64_t                 sellValuesSize,
                        int64_t                 sliceSize,
                        void*                   sellSliceOffsets,
                        void*                   sellColInd,
                        void*                   sellValues,
                        cusparseIndexType_t     sellSliceOffsetsType,
                        cusparseIndexType_t     sellColIndType,
                        cusparseIndexBase_t     idxBase,
                        cudaDataType            valueType)

cusparseStatus_t
cusparseCreateConstSlicedEll(cusparseConstSpMatDescr_t* spMatDescr,
                             int64_t                    rows,
                             int64_t                    cols,
                             int64_t                    nnz,
                             int64_t                    sellValuesSize,
                             int64_t                    sliceSize,
                             const void*                sellSliceOffsets,
                             const void*                sellColInd,
                             const void*                sellValues,
                             cusparseIndexType_t        sellSliceOffsetsType,
                             cusparseIndexType_t        sellColIndType,
                             cusparseIndexBase_t        idxBase,
                             cudaDataType               valueType)

This function initializes the sparse matrix descriptor spMatDescr for the Sliced Ellpack (SELL) format.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	OUT	Sparse matrix descriptor
`rows`	HOST	IN	Number of rows of the sparse matrix
`cols`	HOST	IN	Number of columns of the sparse matrix
`nnz`	HOST	IN	Number of nonzero elements in the sparse matrix
`sellValuesSize`	HOST	IN	Total number of elements in `sellValues` array (nonzero and padding)
`sliceSize`	HOST	IN	The number of rows per slice
`sellSliceOffsets`	DEVICE	IN	Slice offsets of the sparse matrix. Array of size \(\left \lceil{\frac{rows}{sliceSize}}\right \rceil + 1\)
`sellColInd`	DEVICE	IN	Column indexes of the sparse matrix. Array of size `sellValuesSize`
`sellValues`	DEVICE	IN	Values of the sparse matrix. Array of size `sellValuesSize` elements
`sellSliceOffsetsType`	HOST	IN	Data type of `sellSliceOffsets`
`sellColIndType`	HOST	IN	Data type of `sellColInd`
`idxBase`	HOST	IN	Index base of `sellColInd`
`valueType`	HOST	IN	Data type of `sellValues`

Note

Sliced Ellpack Column array sellColInd contains -1 values for indicating padded entries.

cusparseCreateSlicedEll() has the following constraints:

sellSliceOffsets, sellColInd, and sellValues must be aligned to the size of the datatypes specified by sellSliceOffsetsType, sellColIndType, and valueType, respectively. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

Block Sparse Row (BSR)

cusparseCreateBsr()

cusparseStatus_t
cusparseCreateBsr(cusparseSpMatDescr_t* spMatDescr,
                  int64_t               brows,
                  int64_t               bcols,
                  int64_t               bnnz,
                  int64_t               rowBlockSize,
                  int64_t               colBlockSize,
                  void*                 bsrRowOffsets,
                  void*                 bsrColInd,
                  void*                 bsrValues,
                  cusparseIndexType_t   bsrRowOffsetsType,
                  cusparseIndexType_t   bsrColIndType,
                  cusparseIndexBase_t   idxBase,
                  cudaDataType          valueType,
                  cusparseOrder_t       order)

cusparseStatus_t
cusparseCreateConstBsr(cusparseConstSpMatDescr_t* spMatDescr,
                       int64_t                    brows,
                       int64_t                    bcols,
                       int64_t                    bnnz,
                       int64_t                    rowBlockSize,
                       int64_t                    colBlockSize,
                       const void*                bsrRowOffsets,
                       const void*                bsrColInd,
                       const void*                bsrValues,
                       cusparseIndexType_t        bsrRowOffsetsType,
                       cusparseIndexType_t        bsrColIndType,
                       cusparseIndexBase_t        idxBase,
                       cudaDataType               valueType,
                       cusparseOrder_t            order)

This function initializes the sparse matrix descriptor spMatDescr for the Block Compressed Row (BSR) format.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	OUT	Sparse matrix descriptor
`brows`	HOST	IN	Number of block rows of the sparse matrix
`bcols`	HOST	IN	Number of block columns of the sparse matrix
`bnnz`	HOST	IN	Number of blocks of the sparse matrix
`rowBlockSize`	HOST	IN	Number of rows of each block
`colBlockSize`	HOST	IN	Number of columns of each block
`bsrRowOffsets`	DEVICE	IN	Block row offsets of the sparse matrix. Array of size `brows + 1`
`bsrColInd`	DEVICE	IN	Block column indices of the sparse matrix. Array of size `bnnz`
`bsrValues`	DEVICE	IN	Values of the sparse matrix. Array of size `bnnz * rowBlockSize * colBlockSize`
`bsrRowOffsetsType`	HOST	IN	Data type of `bsrRowOffsets`
`bsrColIndType`	HOST	IN	Data type of `bsrColInd`
`idxBase`	HOST	IN	Base index of `bsrRowOffsets` and `bsrColInd`
`valueType`	HOST	IN	Datatype of `bsrValues`
`order`	HOST	IN	Enumerator specifying the memory layout of values in each block

cusparseCreateBsr() has the following constraints:

bsrRowOffsets, bsrColInd, and bsrValues must be aligned to the size of the datatypes specified by bsrRowOffsetsType, bsrColIndType, and valueType, respectively. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

cusparseBsrSetStridedBatch()

cusparseStatus_t
cusparseBsrSetStridedBatch(cusparseSpMatDescr_t spMatDescr,
                           int                  batchCount,
                           int64_t              offsetsBatchStride,
                           int64_t              columnsBatchStride,
                           int64_t              valuesBatchStride)

This function sets the batchCount and the batchStride fields of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`batchCount`	HOST	IN	Number of batches of the sparse matrix
`offsetsBatchStride`	HOST	IN	Address offset between consecutive batches for the row offset array
`columnsBatchStride`	HOST	IN	Address offset between consecutive batches for the column array
`valuesBatchStride`	HOST	IN	Address offset between consecutive batches for the values array

See cusparseStatus_t for the description of the return status.

All Sparse Formats

cusparseDestroySpMat()

cusparseStatus_t
cusparseDestroySpMat(cusparseConstSpMatDescr_t spMatDescr) // non-const descriptor supported

This function releases the host memory allocated for the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor

See cusparseStatus_t for the description of the return status.

cusparseSpMatGetSize()

cusparseStatus_t
cusparseSpMatGetSize(cusparseConstSpMatDescr_t spMatDescr,  // non-const descriptor supported
                     int64_t*                  rows,
                     int64_t*                  cols,
                     int64_t*                  nnz)

This function returns the sizes of the sparse matrix spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`rows`	HOST	OUT	Number of rows of the sparse matrix
`cols`	HOST	OUT	Number of columns of the sparse matrix
`nnz`	HOST	OUT	Number of non-zero entries of the sparse matrix

See cusparseStatus_t for the description of the return status.

cusparseSpMatGetFormat()

cusparseStatus_t
cusparseSpMatGetFormat(cusparseConstSpMatDescr_t spMatDescr, // non-const descriptor supported
                       cusparseFormat_t*         format)

This function returns the format field of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`format`	HOST	OUT	Storage format of the sparse matrix

See cusparseStatus_t for the description of the return status.

cusparseSpMatGetIndexBase()

cusparseStatus_t
cusparseSpMatGetIndexBase(cusparseConstSpMatDescr_t spMatDescr, // non-const descriptor supported
                          cusparseIndexBase_t*      idxBase)

This function returns the idxBase field of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`idxBase`	HOST	OUT	Index base of the sparse matrix

See cusparseStatus_t for the description of the return status.

cusparseSpMatGetValues()

cusparseStatus_t
cusparseSpMatGetValues(cusparseSpMatDescr_t spMatDescr,
                       void**               values)

cusparseStatus_t
cusparseConstSpMatGetValues(cusparseConstSpMatDescr_t spMatDescr,
                            const void**              values)

This function returns the values field of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`values`	DEVICE	OUT	Values of the sparse matrix. Array with `nnz` elements

See cusparseStatus_t for the description of the return status.

cusparseSpMatSetValues()

cusparseStatus_t
cusparseSpMatSetValues(cusparseSpMatDescr_t spMatDescr,
                       void*                values)

This function sets the values field of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`values`	DEVICE	IN	Values of the sparse matrix. Array with `nnz` elements

cusparseSpMatSetValues() has the following constraints:

values must be aligned to the size of its corresponding datatype specified in spMatDescr. See cudaDataType_t for the description of the datatypes.

See cusparseStatus_t for the description of the return status.

cusparseSpMatGetStridedBatch()

cusparseStatus_t
cusparseSpMatGetStridedBatch(cusparseConstSpMatDescr_t spMatDescr, // non-const descriptor supported
                             int*                      batchCount)

This function returns the batchCount field of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`batchCount`	HOST	OUT	Number of batches of the sparse matrix

See cusparseStatus_t for the description of the return status.

cusparseSpMatGetAttribute()

cusparseStatus_t
cusparseSpMatGetAttribute(cusparseConstSpMatDescr_t spMatDescr, // non-const descriptor supported
                          cusparseSpMatAttribute_t  attribute,
                          void*                     data,
                          size_t                    dataSize)

The function gets the attributes of the sparse matrix descriptor spMatDescr.

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	IN	Sparse matrix descriptor
`attribute`	HOST	IN	Attribute enumerator
`data`	HOST	OUT	Attribute value
`dataSize`	HOST	IN	Size of the attribute in bytes for safety

Attribute	Meaning	Possible Values
`CUSPARSE_SPMAT_FILL_MODE`	Indicates if the lower or upper part of a matrix is stored in sparse storage	`CUSPARSE_FILL_MODE_LOWER` `CUSPARSE_FILL_MODE_UPPER`
`CUSPARSE_SPMAT_DIAG_TYPE`	Indicates if the matrix diagonal entries are unity	`CUSPARSE_DIAG_TYPE_NON_UNIT` `CUSPARSE_DIAG_TYPE_UNIT`

See cusparseStatus_t for the description of the return status.

cusparseSpMatSetAttribute()

cusparseStatus_t
cusparseSpMatSetAttribute(cusparseSpMatDescr_t     spMatDescr,
                          cusparseSpMatAttribute_t attribute,
                          const void*              data,
                          size_t                   dataSize)

The function sets the attributes of the sparse matrix descriptor spMatDescr

Param.	Memory	In/out	Meaning
`spMatDescr`	HOST	OUT	Sparse matrix descriptor
`attribute`	HOST	IN	Attribute enumerator
`data`	HOST	IN	Attribute value
`dataSize`	HOST	IN	Size of the attribute in bytes for safety

Attribute	Meaning	Possible Values
`CUSPARSE_SPMAT_FILL_MODE`	Indicates if the lower or upper part of a matrix is stored in sparse storage	`CUSPARSE_FILL_MODE_LOWER` `CUSPARSE_FILL_MODE_UPPER`
`CUSPARSE_SPMAT_DIAG_TYPE`	Indicates if the matrix diagonal entries are unity	`CUSPARSE_DIAG_TYPE_NON_UNIT` `CUSPARSE_DIAG_TYPE_UNIT`

See cusparseStatus_t for the description of the return status.