Crate baracuda_cusparse_sys 

Raw FFI + dynamic loader for NVIDIA cuSPARSE (generic API subset).

baracuda-cusparse wraps this with a safe, typed API. Use this crate directly only if you need a function that the safe layer hasn’t wrapped yet (in which case please file a bug).

Structs

Cusparse

Loaded cuSPARSE shared library plus a per-symbol OnceLock of function pointers.

cusparseStatus_t

Status / error code returned by cuSPARSE FFI calls.

Enums

cudaDataType

cudaDataType values used by cuSPARSE / cuSOLVER’s generic APIs. Only the subset we actually use at v0.1.

cusparseCsr2CscAlg_t

Algorithm selector for CSR-to-CSC conversion.

cusparseDiagType_t

Diagonal-unit selector for sparse triangular routines.

cusparseFillMode_t

Triangular fill mode for sparse triangular routines.

cusparseIndexBase_t

Zero- vs one-based indexing for sparse-matrix index arrays.

cusparseIndexType_t

Index-element dtype for sparse-matrix offsets and column indices.

cusparseOperation_t

Transpose selector for cuSPARSE routines (matches cuBLAS values).

cusparseOrder_t

Row-major vs column-major dense storage order.

cusparseSDDMMAlg_t

Algorithm selector for cusparseSDDMM.

cusparseSpGEMMAlg_t

Algorithm selector for cusparseSpGEMM.

cusparseSpMMAlg_t

Algorithm selector for cusparseSpMM.

cusparseSpMVAlg_t

Algorithm selector for cusparseSpMV.

cusparseSpMatAttribute_t

Attribute key for cusparseSpMatSetAttribute / *GetAttribute.

cusparseSpSMAlg_t

Algorithm selector for cusparseSpSM.

cusparseSpSVAlg_t

Algorithm selector for cusparseSpSV.

Functions

cusparse

Lazy-load the cuSPARSE shared library and return its function-pointer table.

Type Aliases

PFN_cusparseAxpby

cuSPARSE: scaled vector addition (alphax + betay → y). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCooSetPointers

cuSPARSE: replace the row/column/value pointers on a COO descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCreate

cuSPARSE: create a cuSPARSE handle. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCreateBsr

cuSPARSE: create a BSR sparse-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCreateCoo

cuSPARSE: create a COO sparse-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCreateCsc

cuSPARSE: create a CSC sparse-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCreateCsr

cuSPARSE: create a CSR sparse-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCreateDnMat

cuSPARSE: create a dense-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCreateDnVec

cuSPARSE: create a dense-vector descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCscSetPointers

cuSPARSE: replace the column/row/value pointers on a CSC descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCsr2cscEx2

cuSPARSE: convert a CSR matrix to CSC (extended API). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCsr2cscEx2_bufferSize

cuSPARSE: workspace-size query for cusparseCsr2cscEx2. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseCsrSetPointers

cuSPARSE: replace the row/column/value pointers on a CSR descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseDenseToSparse_analysis

cuSPARSE: analysis stage for dense-to-sparse conversion. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseDenseToSparse_bufferSize

cuSPARSE: workspace-size query for dense-to-sparse conversion. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseDenseToSparse_convert

cuSPARSE: execute stage for dense-to-sparse conversion. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseDestroy

cuSPARSE: destroy a cuSPARSE handle. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseDestroyDnMat

cuSPARSE: destroy a dense-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseDestroyDnVec

cuSPARSE: destroy a dense-vector descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseDestroySpMat

cuSPARSE: destroy a sparse-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseGather

cuSPARSE: gather dense entries into a sparse vector. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseGetVersion

cuSPARSE: return the cuSPARSE library version. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseRot

cuSPARSE: Givens-rotation on sparse and dense vector pair. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSDDMM

cuSPARSE: sampled dense-dense matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSDDMM_bufferSize

cuSPARSE: workspace-size query for sampled dense-dense matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSDDMM_preprocess

cuSPARSE: preprocess stage of sampled dense-dense matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseScatter

cuSPARSE: scatter sparse vector entries into a dense vector. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSetStream

cuSPARSE: bind a CUDA stream to a cuSPARSE handle. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpGEMM_compute

cuSPARSE: compute stage of sparse matrix × sparse matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpGEMM_copy

cuSPARSE: copy stage of sparse matrix × sparse matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpGEMM_createDescr

cuSPARSE: create an opaque descriptor for sparse matrix × sparse matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpGEMM_destroyDescr

cuSPARSE: destroy an opaque descriptor for sparse matrix × sparse matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpGEMM_workEstimation

cuSPARSE: work-estimation stage of sparse matrix × sparse matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpMM

cuSPARSE: sparse matrix × dense matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpMM_bufferSize

cuSPARSE: workspace-size query for sparse matrix × dense matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpMM_preprocess

cuSPARSE: preprocess stage of sparse matrix × dense matrix multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpMV

cuSPARSE: sparse matrix-vector multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpMV_bufferSize

cuSPARSE: workspace-size query for sparse matrix-vector multiplication. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpMatGetSize

cuSPARSE: query rows, cols, and nnz of a sparse-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpMatSetAttribute

cuSPARSE: set an attribute (fill mode, diag type) on a sparse-matrix descriptor. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSM_analysis

cuSPARSE: analysis stage of sparse triangular linear solve (multiple right-hand sides). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSM_bufferSize

cuSPARSE: workspace-size query for sparse triangular linear solve (multiple right-hand sides). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSM_createDescr

cuSPARSE: create an opaque descriptor for sparse triangular linear solve (multiple right-hand sides). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSM_destroyDescr

cuSPARSE: destroy an opaque descriptor for sparse triangular linear solve (multiple right-hand sides). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSM_solve

cuSPARSE: solve stage of sparse triangular linear solve (multiple right-hand sides). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSV_analysis

cuSPARSE: analysis stage of sparse triangular linear solve (single right-hand side). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSV_bufferSize

cuSPARSE: workspace-size query for sparse triangular linear solve (single right-hand side). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSV_createDescr

cuSPARSE: create an opaque descriptor for sparse triangular linear solve (single right-hand side). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSV_destroyDescr

cuSPARSE: destroy an opaque descriptor for sparse triangular linear solve (single right-hand side). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSpSV_solve

cuSPARSE: solve stage of sparse triangular linear solve (single right-hand side). See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSparseToDense

cuSPARSE: materialize a sparse matrix into a dense one. See https://docs.nvidia.com/cuda/cusparse/index.html.

PFN_cusparseSparseToDense_bufferSize

cuSPARSE: workspace-size query for sparse-to-dense conversion. See https://docs.nvidia.com/cuda/cusparse/index.html.

cusparseDnMatDescr_t

Opaque dense-matrix descriptor.

cusparseDnVecDescr_t

Opaque dense-vector descriptor.

cusparseHandle_t

Opaque cuSPARSE handle.

cusparseMatDescr_t

Opaque legacy-API matrix descriptor.

cusparseSpGEMMDescr_t

Opaque SpGEMM intermediate-state descriptor.

cusparseSpMatDescr_t

Opaque sparse-matrix descriptor (CSR / CSC / COO / BSR).

cusparseSpSMDescr_t

Opaque SpSM intermediate-state descriptor.

cusparseSpSVDescr_t

Opaque SpSV intermediate-state descriptor.