Namespace NVMathNet.Interop

Classes

CuBlasException

Represents an error returned by cuBLAS / cuBLASLt.

CuBlasNative

Raw P/Invoke bindings for cuBLAS and cuBLASLt.

CuDssException

Exception thrown for cuDSS errors.

CuDssNative

Raw bindings for cuDSS (sparse direct solver). Signatures verified against cudss.h 0.8.

CuFftException

Represents an error returned by cuFFT.

CuFftNative

Raw P/Invoke bindings for the cuFFT library.

CuRandException

Represents an error returned by cuRAND.

CuRandNative

Raw P/Invoke bindings for cuRAND.

CuSolverException

Represents an error returned by cuSOLVER.

CuSolverNative

Raw P/Invoke bindings for cuSOLVER dense (cusolverDn).

CuSparseException

Represents an error returned by cuSPARSE.

CuSparseNative

Raw P/Invoke bindings for cuSPARSE.

CuTensorException

Represents an error returned by cuTENSOR.

CuTensorNative

Raw P/Invoke bindings for cuTENSOR v2.

CudaException

Represents an error returned by the CUDA Runtime API.

CudaRuntime

Raw P/Invoke bindings for the CUDA Runtime API (cudart). All methods are loaded lazily via NVMathNet.Interop.NativeLibraryLoader.

NativeLibraryLoader

Cross-platform native library loader using System.Runtime.InteropServices.NativeLibrary so that both Windows (LoadLibrary) and Linux (dlopen) are supported without any additional code.

Enums

CuBlasComputeType

cuBLAS / cuBLASLt compute types.

CuBlasDiagType

Diagonal type for triangular matrices.

CuBlasFillMode

cuBLAS fill mode for symmetric/triangular matrices.

CuBlasLtEpilogue

cuBLASLt epilog operations applied after the matrix multiplication.

CuBlasOperation

cuBLAS operation types (transpose).

CuBlasSideMode

Side mode for cuBLAS structured operations.

CuBlasStatus

cuBLAS status codes.

CuDssConfigParam

Solver configuration parameters (cudssConfigParam_t).

CuDssDataParam

Solver data parameters (cudssDataParam_t).

CuDssDataType

Value types used by cuDSS (cudssDataType_t; same numbering as cudaDataType).

CuDssMatrixType

Structure of the system matrix (cudssMatrixType_t).

CuDssMatrixView

Which part of the matrix is stored (cudssMatrixViewType_t).

CuDssPhase

Solver phases (cudssPhase_t); phases can be combined with bitwise or.

CuDssReorderingAlg

Fill-reducing reordering algorithms (cudssReorderingAlg_t).

CuDssStatus

cuDSS status codes (cudssStatus_t).

CuFftDirection

cuFFT transform direction.

CuFftResult

cuFFT status codes.

CuFftType

cuFFT transform types.

CuRandRngType

cuRAND random number generator types.

CuRandStatus

cuRAND status codes.

CuSolverEigMode

cuSOLVER eigenvalue computation mode.

CuSolverStatus

cuSOLVER status codes.

CuSparseDiagType

Whether the diagonal of a triangular sparse matrix is implicitly all ones.

CuSparseFillMode

Which triangle of a sparse matrix is referenced by a triangular solve.

CuSparseFormat

cuSPARSE sparse matrix format.

CuSparseIndexBase

cuSPARSE index base.

CuSparseIndexType

cuSPARSE index types.

CuSparseOperation

cuSPARSE operation type applied to a sparse or dense matrix operand.

CuSparseSddmmAlg

Algorithm selector for cusparseSDDMM*.

CuSparseSpGemmAlg

Algorithm selector for cusparseSpGEMM_*.

CuSparseSpMMAlg

cuSPARSE SpMM algorithms.

CuSparseSpMVAlg

cuSPARSE SpMV algorithms.

CuSparseSpMatAttribute

Sparse matrix attributes for cusparseSpMatSetAttribute.

CuSparseSpSMAlg

Algorithm selector for cusparseSpSM_*.

CuSparseSpSVAlg

Algorithm selector for cusparseSpSV_*.

CuSparseStatus

cuSPARSE status codes.

CuTensorAlgo

cuTENSOR contraction algorithms.

CuTensorJitMode

cuTENSOR JIT compilation mode.

CuTensorOperator

cuTENSOR operators: unary operators applied to individual operands and binary operators that combine two operands. Values match cutensorOperator_t in cutensor/types.h.

CuTensorStatus

cuTENSOR status codes.

CuTensorWorksizePreference

cuTENSOR workspace size preference hint.

CudaDataType

CUDA data types used across cuBLASLt, cuSPARSE and cuTENSOR.

CudaError

CUDA error codes as returned by the CUDA Runtime API.

CudaMemcpyKind

Specifies the direction of a memory copy operation.