nvmath-python 1.0.0__cp314-cp314t-win_amd64.whl

nvmath/bindings/_internal/cusparseLt.pxd

@@ -0,0 +1,48 @@
+# Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. ALL RIGHTS RESERVED.
+#
+# SPDX-License-Identifier: Apache-2.0
+#
+# This code was automatically generated across versions from 0.7.1 to 0.8.1, generator version 0.3.1.dev1565+g7fa82f8eb. Do not modify it directly.
+
+from libc.stdint cimport int64_t, uint8_t, uint32_t, int32_t
+from ..cycusparseLt cimport *
+from ..cycusparseLt cimport cudaDataType, cudaStream_t, libraryPropertyType, cusparseComputeType
+from ..cycusparse cimport cusparseStatus_t, _CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR, cusparseOrder_t, cusparseOperation_t
+
+cdef void* load_library(const int driver_ver) except* with gil
+
+###############################################################################
+# Wrapper functions
+###############################################################################
+
+cdef const char* _cusparseLtGetErrorName(cusparseStatus_t status) except?NULL nogil
+cdef const char* _cusparseLtGetErrorString(cusparseStatus_t status) except?NULL nogil
+cdef cusparseStatus_t _cusparseLtInit(cusparseLtHandle_t* handle) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtDestroy(const cusparseLtHandle_t* handle) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtGetVersion(const cusparseLtHandle_t* handle, int* version) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtGetProperty(libraryPropertyType propertyType, int* value) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtDenseDescriptorInit(const cusparseLtHandle_t* handle, cusparseLtMatDescriptor_t* matDescr, int64_t rows, int64_t cols, int64_t ld, uint32_t alignment, cudaDataType valueType, cusparseOrder_t order) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtStructuredDescriptorInit(const cusparseLtHandle_t* handle, cusparseLtMatDescriptor_t* matDescr, int64_t rows, int64_t cols, int64_t ld, uint32_t alignment, cudaDataType valueType, cusparseOrder_t order, cusparseLtSparsity_t sparsity) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatDescriptorDestroy(const cusparseLtMatDescriptor_t* matDescr) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatDescSetAttribute(const cusparseLtHandle_t* handle, cusparseLtMatDescriptor_t* matmulDescr, cusparseLtMatDescAttribute_t matAttribute, const void* data, size_t dataSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatDescGetAttribute(const cusparseLtHandle_t* handle, const cusparseLtMatDescriptor_t* matmulDescr, cusparseLtMatDescAttribute_t matAttribute, void* data, size_t dataSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulDescriptorInit(const cusparseLtHandle_t* handle, cusparseLtMatmulDescriptor_t* matmulDescr, cusparseOperation_t opA, cusparseOperation_t opB, const cusparseLtMatDescriptor_t* matA, const cusparseLtMatDescriptor_t* matB, const cusparseLtMatDescriptor_t* matC, const cusparseLtMatDescriptor_t* matD, cusparseComputeType computeType) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulDescSetAttribute(const cusparseLtHandle_t* handle, cusparseLtMatmulDescriptor_t* matmulDescr, cusparseLtMatmulDescAttribute_t matmulAttribute, const void* data, size_t dataSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulDescGetAttribute(const cusparseLtHandle_t* handle, const cusparseLtMatmulDescriptor_t* matmulDescr, cusparseLtMatmulDescAttribute_t matmulAttribute, void* data, size_t dataSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulAlgSelectionInit(const cusparseLtHandle_t* handle, cusparseLtMatmulAlgSelection_t* algSelection, const cusparseLtMatmulDescriptor_t* matmulDescr, cusparseLtMatmulAlg_t alg) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulAlgSetAttribute(const cusparseLtHandle_t* handle, cusparseLtMatmulAlgSelection_t* algSelection, cusparseLtMatmulAlgAttribute_t attribute, const void* data, size_t dataSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulAlgGetAttribute(const cusparseLtHandle_t* handle, const cusparseLtMatmulAlgSelection_t* algSelection, cusparseLtMatmulAlgAttribute_t attribute, void* data, size_t dataSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulGetWorkspace(const cusparseLtHandle_t* handle, const cusparseLtMatmulPlan_t* plan, size_t* workspaceSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulPlanInit(const cusparseLtHandle_t* handle, cusparseLtMatmulPlan_t* plan, const cusparseLtMatmulDescriptor_t* matmulDescr, const cusparseLtMatmulAlgSelection_t* algSelection) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulPlanDestroy(const cusparseLtMatmulPlan_t* plan) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmul(const cusparseLtHandle_t* handle, const cusparseLtMatmulPlan_t* plan, const void* alpha, const void* d_A, const void* d_B, const void* beta, const void* d_C, void* d_D, void* workspace, cudaStream_t* streams, int32_t numStreams) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulSearch(const cusparseLtHandle_t* handle, cusparseLtMatmulPlan_t* plan, const void* alpha, const void* d_A, const void* d_B, const void* beta, const void* d_C, void* d_D, void* workspace, cudaStream_t* streams, int32_t numStreams) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtSpMMAPrune(const cusparseLtHandle_t* handle, const cusparseLtMatmulDescriptor_t* matmulDescr, const void* d_in, void* d_out, cusparseLtPruneAlg_t pruneAlg, cudaStream_t stream) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtSpMMAPruneCheck(const cusparseLtHandle_t* handle, const cusparseLtMatmulDescriptor_t* matmulDescr, const void* d_in, int* valid, cudaStream_t stream) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtSpMMAPrune2(const cusparseLtHandle_t* handle, const cusparseLtMatDescriptor_t* sparseMatDescr, int isSparseA, cusparseOperation_t op, const void* d_in, void* d_out, cusparseLtPruneAlg_t pruneAlg, cudaStream_t stream) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtSpMMAPruneCheck2(const cusparseLtHandle_t* handle, const cusparseLtMatDescriptor_t* sparseMatDescr, int isSparseA, cusparseOperation_t op, const void* d_in, int* d_valid, cudaStream_t stream) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtSpMMACompressedSize(const cusparseLtHandle_t* handle, const cusparseLtMatmulPlan_t* plan, size_t* compressedSize, size_t* compressedBufferSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtSpMMACompress(const cusparseLtHandle_t* handle, const cusparseLtMatmulPlan_t* plan, const void* d_dense, void* d_compressed, void* d_compressed_buffer, cudaStream_t stream) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtSpMMACompressedSize2(const cusparseLtHandle_t* handle, const cusparseLtMatDescriptor_t* sparseMatDescr, size_t* compressedSize, size_t* compressedBufferSize) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtSpMMACompress2(const cusparseLtHandle_t* handle, const cusparseLtMatDescriptor_t* sparseMatDescr, int isSparseA, cusparseOperation_t op, const void* d_dense, void* d_compressed, void* d_compressed_buffer, cudaStream_t stream) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cusparseStatus_t _cusparseLtMatmulAlgSelectionDestroy(const cusparseLtMatmulAlgSelection_t* algSelection) except?_CUSPARSESTATUS_T_INTERNAL_LOADING_ERROR nogil

nvmath/bindings/_internal/cutensor.pxd

@@ -0,0 +1,58 @@
+# Copyright (c) 2025-2026, NVIDIA CORPORATION & AFFILIATES. ALL RIGHTS RESERVED.
+#
+# SPDX-License-Identifier: Apache-2.0
+#
+# This code was automatically generated with version 2.5.0, generator version 0.3.1.dev1393+g0a20dc9d7. Do not modify it directly.
+
+from ..cycutensor cimport *
+
+
+###############################################################################
+# Wrapper functions
+###############################################################################
+
+cdef cutensorStatus_t _cutensorCreate(cutensorHandle_t* handle) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorDestroy(cutensorHandle_t handle) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorHandleResizePlanCache(cutensorHandle_t handle, const uint32_t numEntries) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorHandleWritePlanCacheToFile(const cutensorHandle_t handle, const char filename[]) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorHandleReadPlanCacheFromFile(cutensorHandle_t handle, const char filename[], uint32_t* numCachelinesRead) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorWriteKernelCacheToFile(const cutensorHandle_t handle, const char filename[]) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorReadKernelCacheFromFile(cutensorHandle_t handle, const char filename[]) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreateTensorDescriptor(const cutensorHandle_t handle, cutensorTensorDescriptor_t* desc, const uint32_t numModes, const int64_t extent[], const int64_t stride[], cudaDataType_t dataType, uint32_t alignmentRequirement) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorDestroyTensorDescriptor(cutensorTensorDescriptor_t desc) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreateElementwiseTrinary(const cutensorHandle_t handle, cutensorOperationDescriptor_t* desc, const cutensorTensorDescriptor_t descA, const int32_t modeA[], cutensorOperator_t opA, const cutensorTensorDescriptor_t descB, const int32_t modeB[], cutensorOperator_t opB, const cutensorTensorDescriptor_t descC, const int32_t modeC[], cutensorOperator_t opC, const cutensorTensorDescriptor_t descD, const int32_t modeD[], cutensorOperator_t opAB, cutensorOperator_t opABC, const cutensorComputeDescriptor_t descCompute) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorElementwiseTrinaryExecute(const cutensorHandle_t handle, const cutensorPlan_t plan, const void* alpha, const void* A, const void* beta, const void* B, const void* gamma, const void* C, void* D, cudaStream_t stream) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreateElementwiseBinary(const cutensorHandle_t handle, cutensorOperationDescriptor_t* desc, const cutensorTensorDescriptor_t descA, const int32_t modeA[], cutensorOperator_t opA, const cutensorTensorDescriptor_t descC, const int32_t modeC[], cutensorOperator_t opC, const cutensorTensorDescriptor_t descD, const int32_t modeD[], cutensorOperator_t opAC, const cutensorComputeDescriptor_t descCompute) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorElementwiseBinaryExecute(const cutensorHandle_t handle, const cutensorPlan_t plan, const void* alpha, const void* A, const void* gamma, const void* C, void* D, cudaStream_t stream) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreatePermutation(const cutensorHandle_t handle, cutensorOperationDescriptor_t* desc, const cutensorTensorDescriptor_t descA, const int32_t modeA[], cutensorOperator_t opA, const cutensorTensorDescriptor_t descB, const int32_t modeB[], const cutensorComputeDescriptor_t descCompute) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorPermute(const cutensorHandle_t handle, const cutensorPlan_t plan, const void* alpha, const void* A, void* B, const cudaStream_t stream) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreateContraction(const cutensorHandle_t handle, cutensorOperationDescriptor_t* desc, const cutensorTensorDescriptor_t descA, const int32_t modeA[], cutensorOperator_t opA, const cutensorTensorDescriptor_t descB, const int32_t modeB[], cutensorOperator_t opB, const cutensorTensorDescriptor_t descC, const int32_t modeC[], cutensorOperator_t opC, const cutensorTensorDescriptor_t descD, const int32_t modeD[], const cutensorComputeDescriptor_t descCompute) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorDestroyOperationDescriptor(cutensorOperationDescriptor_t desc) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorOperationDescriptorSetAttribute(const cutensorHandle_t handle, cutensorOperationDescriptor_t desc, cutensorOperationDescriptorAttribute_t attr, const void* buf, size_t sizeInBytes) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorOperationDescriptorGetAttribute(const cutensorHandle_t handle, cutensorOperationDescriptor_t desc, cutensorOperationDescriptorAttribute_t attr, void* buf, size_t sizeInBytes) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreatePlanPreference(const cutensorHandle_t handle, cutensorPlanPreference_t* pref, cutensorAlgo_t algo, cutensorJitMode_t jitMode) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorDestroyPlanPreference(cutensorPlanPreference_t pref) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorPlanPreferenceGetAttribute(const cutensorHandle_t handle, cutensorPlanPreference_t pref, cutensorPlanPreferenceAttribute_t attr, void* buf, size_t sizeInBytes) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorPlanPreferenceSetAttribute(const cutensorHandle_t handle, cutensorPlanPreference_t pref, cutensorPlanPreferenceAttribute_t attr, const void* buf, size_t sizeInBytes) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorPlanGetAttribute(const cutensorHandle_t handle, const cutensorPlan_t plan, cutensorPlanAttribute_t attr, void* buf, size_t sizeInBytes) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorEstimateWorkspaceSize(const cutensorHandle_t handle, const cutensorOperationDescriptor_t desc, const cutensorPlanPreference_t planPref, const cutensorWorksizePreference_t workspacePref, uint64_t* workspaceSizeEstimate) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreatePlan(const cutensorHandle_t handle, cutensorPlan_t* plan, const cutensorOperationDescriptor_t desc, const cutensorPlanPreference_t pref, uint64_t workspaceSizeLimit) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorDestroyPlan(cutensorPlan_t plan) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorContract(const cutensorHandle_t handle, const cutensorPlan_t plan, const void* alpha, const void* A, const void* B, const void* beta, const void* C, void* D, void* workspace, uint64_t workspaceSize, cudaStream_t stream) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreateReduction(const cutensorHandle_t handle, cutensorOperationDescriptor_t* desc, const cutensorTensorDescriptor_t descA, const int32_t modeA[], cutensorOperator_t opA, const cutensorTensorDescriptor_t descC, const int32_t modeC[], cutensorOperator_t opC, const cutensorTensorDescriptor_t descD, const int32_t modeD[], cutensorOperator_t opReduce, const cutensorComputeDescriptor_t descCompute) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorReduce(const cutensorHandle_t handle, const cutensorPlan_t plan, const void* alpha, const void* A, const void* beta, const void* C, void* D, void* workspace, uint64_t workspaceSize, cudaStream_t stream) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreateContractionTrinary(const cutensorHandle_t handle, cutensorOperationDescriptor_t* desc, const cutensorTensorDescriptor_t descA, const int32_t modeA[], cutensorOperator_t opA, const cutensorTensorDescriptor_t descB, const int32_t modeB[], cutensorOperator_t opB, const cutensorTensorDescriptor_t descC, const int32_t modeC[], cutensorOperator_t opC, const cutensorTensorDescriptor_t descD, const int32_t modeD[], cutensorOperator_t opD, const cutensorTensorDescriptor_t descE, const int32_t modeE[], const cutensorComputeDescriptor_t descCompute) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorContractTrinary(const cutensorHandle_t handle, const cutensorPlan_t plan, const void* alpha, const void* A, const void* B, const void* C, const void* beta, const void* D, void* E, void* workspace, uint64_t workspaceSize, cudaStream_t stream) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreateBlockSparseTensorDescriptor(cutensorHandle_t handle, cutensorBlockSparseTensorDescriptor_t* desc, const uint32_t numModes, const uint64_t numNonZeroBlocks, const uint32_t numSectionsPerMode[], const int64_t extent[], const int32_t nonZeroCoordinates[], const int64_t stride[], cudaDataType_t dataType) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorDestroyBlockSparseTensorDescriptor(cutensorBlockSparseTensorDescriptor_t desc) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorCreateBlockSparseContraction(const cutensorHandle_t handle, cutensorOperationDescriptor_t* desc, const cutensorBlockSparseTensorDescriptor_t descA, const int32_t modeA[], cutensorOperator_t opA, const cutensorBlockSparseTensorDescriptor_t descB, const int32_t modeB[], cutensorOperator_t opB, const cutensorBlockSparseTensorDescriptor_t descC, const int32_t modeC[], cutensorOperator_t opC, const cutensorBlockSparseTensorDescriptor_t descD, const int32_t modeD[], const cutensorComputeDescriptor_t descCompute) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorBlockSparseContract(const cutensorHandle_t handle, const cutensorPlan_t plan, const void* alpha, const void* const A[], const void* const B[], const void* beta, const void* const C[], void* const D[], void* workspace, uint64_t workspaceSize, cudaStream_t stream) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef const char* _cutensorGetErrorString(const cutensorStatus_t error) except?NULL nogil
+cdef size_t _cutensorGetVersion() except?0 nogil
+cdef size_t _cutensorGetCudartVersion() except?0 nogil
+cdef cutensorStatus_t _cutensorLoggerSetCallback(cutensorLoggerCallback_t callback) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorLoggerSetFile(FILE* file) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorLoggerOpenFile(const char* logFile) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorLoggerSetLevel(int32_t level) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorLoggerSetMask(int32_t mask) except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil
+cdef cutensorStatus_t _cutensorLoggerForceDisable() except?_CUTENSORSTATUS_T_INTERNAL_LOADING_ERROR nogil

nvmath/bindings/_internal/mathdx.pxd

@@ -0,0 +1,116 @@
+# This code was automatically generated across versions from 0.3.1 to 0.3.2, generator version 0.3.1.dev1418+g63712a33a.d20260318. Do not modify it directly.
+
+from ..cymathdx cimport *
+
+
+###############################################################################
+# Wrapper functions
+###############################################################################
+
+cdef commondxStatusType _commondxCreateCode(commondxCode* code) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxSetCodeOptionInt64(commondxCode code, commondxOption option, long long int value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxSetCodeOptionInt64s(commondxCode code, commondxOption option, size_t count, long long int* values) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxSetCodeOptionStr(commondxCode code, commondxOption option, const char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetCodeOptionInt64(commondxCode code, commondxOption option, long long int* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetCodeOptionsInt64s(commondxCode code, commondxOption option, size_t size, long long int* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetCodeLTOIRSize(commondxCode code, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetCodeLTOIR(commondxCode code, size_t size, void* out) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetCodeNumLTOIRs(commondxCode code, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetCodeLTOIRSizes(commondxCode code, size_t size, size_t* out) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetCodeLTOIRs(commondxCode code, size_t size, void** out) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxDestroyCode(commondxCode code) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef const char* _commondxStatusToStr(commondxStatusType status) except?NULL nogil
+cdef commondxStatusType _commondxGetLastErrorStrSize(size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetLastErrorStr(commondxStatusType* code, size_t size, char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _mathdxGetVersion(int* version) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _mathdxGetVersionEx(int* major, int* minor, int* patch) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxCreateDescriptor(cublasdxDescriptor* handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxSetOptionStr(cublasdxDescriptor handle, commondxOption option, const char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxSetOperatorInt64(cublasdxDescriptor handle, cublasdxOperatorType op, long long int value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxSetOperatorInt64s(cublasdxDescriptor handle, cublasdxOperatorType op, size_t count, const long long int* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxCreateTensor(cublasdxDescriptor handle, cublasdxTensorType tensor_type, cublasdxTensor* tensor) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxCreateTensorStrided(cublasdxMemorySpace memory_space, commondxValueType value_type, void* ptr, long long int rank, long long int* shape, long long int* stride, cublasdxTensor* tensor) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxMakeTensorLike(cublasdxTensor input, commondxValueType value_type, cublasdxTensor* output) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxDestroyTensor(cublasdxTensor tensor) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxDestroyPipeline(cublasdxPipeline pipeline) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxCreateDevicePipeline(cublasdxDescriptor handle, cublasdxDevicePipelineType device_pipeline_type, long long int pipeline_depth, cublasdxBlockSizeStrategy block_size_strategy, cublasdxTensor tensor_a, cublasdxTensor tensor_b, cublasdxPipeline* device_pipeline) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxCreateTilePipeline(cublasdxDescriptor handle, cublasdxTilePipelineType tile_pipeline_type, cublasdxPipeline device_pipeline, cublasdxPipeline* tile_pipeline) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxSetTensorOptionInt64(cublasdxTensor tensor, cublasdxTensorOption option, long long int value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxFinalizeTensors(size_t count, const cublasdxTensor* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxFinalizePipelines(size_t count, const cublasdxPipeline* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxFinalize(size_t countTensors, const cublasdxTensor* tensors, size_t countPipelines, const cublasdxPipeline* pipelines) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetTensorTraitInt64(cublasdxTensor tensor, cublasdxTensorTrait trait, long long int* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetPipelineTraitInt64(cublasdxPipeline pipeline, cublasdxPipelineTrait trait, long long int* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetPipelineTraitInt64s(cublasdxPipeline pipeline, cublasdxPipelineTrait trait, size_t count, long long int* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetTensorTraitStrSize(cublasdxTensor tensor, cublasdxTensorTrait trait, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetPipelineTraitStrSize(cublasdxPipeline pipeline, cublasdxPipelineTrait trait, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetTensorTraitStr(cublasdxTensor tensor, cublasdxTensorTrait trait, size_t size, char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetPipelineTraitStr(cublasdxPipeline pipeline, cublasdxPipelineTrait trait, size_t size, char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxCreateDeviceFunction(cublasdxDescriptor handle, cublasdxDeviceFunctionType device_function_type, size_t count, const cublasdxTensor* array, cublasdxDeviceFunction* device_function) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxDestroyDeviceFunction(cublasdxDeviceFunction device_function) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxCreateDeviceFunctionWithPipelines(cublasdxDescriptor handle, cublasdxDeviceFunctionType device_function_type, size_t tensor_count, const cublasdxTensor* tensors, size_t pipeline_count, const cublasdxPipeline* pipelines, cublasdxDeviceFunction* device_function) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxFinalizeDeviceFunctions(commondxCode code, size_t count, const cublasdxDeviceFunction* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetDeviceFunctionTraitStrSize(cublasdxDeviceFunction device_function, cublasdxDeviceFunctionTrait trait, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetDeviceFunctionTraitStr(cublasdxDeviceFunction device_function, cublasdxDeviceFunctionTrait trait, size_t size, char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetLTOIRSize(cublasdxDescriptor handle, size_t* lto_size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetLTOIR(cublasdxDescriptor handle, size_t size, void* lto) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetTraitStrSize(cublasdxDescriptor handle, cublasdxTraitType trait, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetTraitStr(cublasdxDescriptor handle, cublasdxTraitType trait, size_t size, char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetTraitInt64(cublasdxDescriptor handle, cublasdxTraitType trait, long long int* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxGetTraitInt64s(cublasdxDescriptor handle, cublasdxTraitType trait, size_t count, long long int* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef const char* _cublasdxOperatorTypeToStr(cublasdxOperatorType op) except?NULL nogil
+cdef const char* _cublasdxTraitTypeToStr(cublasdxTraitType trait) except?NULL nogil
+cdef commondxStatusType _cublasdxFinalizeCode(commondxCode code, cublasdxDescriptor handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cublasdxDestroyDescriptor(cublasdxDescriptor handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxCreateDescriptor(cufftdxDescriptor* handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxSetOptionStr(cufftdxDescriptor handle, commondxOption opt, const char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetKnobInt64Size(cufftdxDescriptor handle, size_t num_knobs, cufftdxKnobType* knobs_ptr, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetKnobInt64s(cufftdxDescriptor handle, size_t num_knobs, cufftdxKnobType* knobs_ptr, size_t size, long long int* values) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxSetOperatorInt64(cufftdxDescriptor handle, cufftdxOperatorType op, long long int value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxSetOperatorInt64s(cufftdxDescriptor handle, cufftdxOperatorType op, size_t count, const long long int* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetLTOIRSize(cufftdxDescriptor handle, size_t* lto_size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetLTOIR(cufftdxDescriptor handle, size_t size, void* lto) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetTraitStrSize(cufftdxDescriptor handle, cufftdxTraitType trait, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetTraitStr(cufftdxDescriptor handle, cufftdxTraitType trait, size_t size, char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetTraitInt64(cufftdxDescriptor handle, cufftdxTraitType trait, long long int* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetTraitInt64s(cufftdxDescriptor handle, cufftdxTraitType trait, size_t count, long long int* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxGetTraitCommondxDataType(cufftdxDescriptor handle, cufftdxTraitType trait, commondxValueType* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxFinalizeCode(commondxCode code, cufftdxDescriptor handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cufftdxDestroyDescriptor(cufftdxDescriptor handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef const char* _cufftdxOperatorTypeToStr(cufftdxOperatorType op) except?NULL nogil
+cdef const char* _cufftdxTraitTypeToStr(cufftdxTraitType op) except?NULL nogil
+cdef commondxStatusType _cusolverdxCreateDescriptor(cusolverdxDescriptor* handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxSetOptionStr(cusolverdxDescriptor handle, commondxOption opt, const char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxSetOperatorInt64(cusolverdxDescriptor handle, cusolverdxOperatorType op, long long int value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxSetOperatorInt64s(cusolverdxDescriptor handle, cusolverdxOperatorType op, size_t count, const long long int* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetLTOIRSize(cusolverdxDescriptor handle, size_t* lto_size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetLTOIR(cusolverdxDescriptor handle, size_t size, void* lto) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetUniversalFATBINSize(cusolverdxDescriptor handle, size_t* fatbin_size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetUniversalFATBIN(cusolverdxDescriptor handle, size_t fatbin_size, void* fatbin) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetTraitStrSize(cusolverdxDescriptor handle, cusolverdxTraitType trait, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetTraitStr(cusolverdxDescriptor handle, cusolverdxTraitType trait, size_t size, char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetTraitInt64(cusolverdxDescriptor handle, cusolverdxTraitType trait, long long int* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetTraitInt64s(cusolverdxDescriptor handle, cusolverdxTraitType trait, size_t count, long long int* values) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxFinalizeCode(commondxCode code, cusolverdxDescriptor handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxDestroyDescriptor(cusolverdxDescriptor handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef const char* _cusolverdxOperatorTypeToStr(cusolverdxOperatorType op) except?NULL nogil
+cdef const char* _cusolverdxTraitTypeToStr(cusolverdxTraitType trait) except?NULL nogil
+cdef commondxStatusType _curanddxGetVersion(int* major, int* minor, int* patch) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxCreateDescriptor(curanddxDescriptor* handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxSetOptionStr(curanddxDescriptor handle, commondxOption opt, const char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxSetOperatorInt64(curanddxDescriptor handle, curanddxOperatorType op, long long int value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxSetOperatorDoubles(curanddxDescriptor handle, curanddxOperatorType op, size_t count, double* values) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxGetTraitStrSize(curanddxDescriptor handle, curanddxTraitType trait, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxGetTraitStr(curanddxDescriptor handle, curanddxTraitType trait, size_t size, char* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxGetTraitInt64(curanddxDescriptor handle, curanddxTraitType trait, long long int* value) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxFinalizeCode(commondxCode code, curanddxDescriptor handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _curanddxDestroyDescriptor(curanddxDescriptor handle) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef const char* _curanddxOperatorTypeToStr(curanddxOperatorType op) except?NULL nogil
+cdef const char* _curanddxDistributionToStr(curanddxDistribution dist) except?NULL nogil
+cdef const char* _curanddxGeneratorToStr(curanddxGenerator generator) except?NULL nogil
+cdef const char* _curanddxGenerateMethodToStr(curanddxGenerateMethod generate_method) except?NULL nogil
+cdef const char* _curanddxNormalMethodToStr(curanddxNormalMethod normal_method) except?NULL nogil
+cdef const char* _curanddxTraitTypeToStr(curanddxTraitType trait) except?NULL nogil
+cdef commondxStatusType _commondxGetCodePTXSize(commondxCode code, size_t* size) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _commondxGetCodePTX(commondxCode code, size_t size, void* out) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil
+cdef commondxStatusType _cusolverdxGetTraitCommondxDataTypes(cusolverdxDescriptor handle, cusolverdxTraitType trait, size_t count, commondxValueType* array) except?_COMMONDXSTATUSTYPE_INTERNAL_LOADING_ERROR nogil

nvmath/bindings/_internal/nvshmem.pxd

@@ -0,0 +1,29 @@
+# Copyright (c) 2025-2026, NVIDIA CORPORATION & AFFILIATES. ALL RIGHTS RESERVED.
+#
+# SPDX-License-Identifier: Apache-2.0
+#
+# This code was automatically generated with version 3.1.7. Do not modify it directly.
+
+from ..cynvshmem cimport *
+
+
+###############################################################################
+# Wrapper functions
+###############################################################################
+
+cdef int _nvshmemx_init_status() except?-42 nogil
+cdef int _nvshmem_my_pe() except?-42 nogil
+cdef int _nvshmem_n_pes() except?-42 nogil
+cdef void* _nvshmem_malloc(size_t size) except?NULL nogil
+cdef void* _nvshmem_calloc(size_t count, size_t size) except?NULL nogil
+cdef void* _nvshmem_align(size_t alignment, size_t size) except?NULL nogil
+cdef void _nvshmem_free(void* ptr) except* nogil
+cdef void* _nvshmem_ptr(const void* dest, int pe) except?NULL nogil
+cdef void _nvshmem_int_p(int* dest, const int value, int pe) except* nogil
+cdef int _nvshmem_team_my_pe(nvshmem_team_t team) except?-42 nogil
+cdef void _nvshmemx_barrier_all_on_stream(cudaStream_t stream) except* nogil
+cdef void _nvshmemx_sync_all_on_stream(cudaStream_t stream) except* nogil
+cdef int _nvshmemx_hostlib_init_attr(unsigned int flags, nvshmemx_init_attr_t* attr) except?-42 nogil
+cdef void _nvshmemx_hostlib_finalize() except* nogil
+cdef int _nvshmemx_set_attr_uniqueid_args(const int myrank, const int nranks, const nvshmemx_uniqueid_t* uniqueid, nvshmemx_init_attr_t* attr) except?-42 nogil
+cdef int _nvshmemx_get_uniqueid(nvshmemx_uniqueid_t* uniqueid) except?-42 nogil

nvmath/bindings/_internal/utils.pxd

@@ -0,0 +1,174 @@
+# Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. ALL RIGHTS RESERVED.
+#
+# SPDX-License-Identifier: Apache-2.0
+
+from libc.stdint cimport int32_t, int64_t, intptr_t, uint32_t
+from libcpp.vector cimport vector
+from libcpp cimport bool as cppbool
+from libcpp cimport nullptr_t, nullptr
+from libcpp.memory cimport unique_ptr
+from .common_types cimport cuComplex, cuDoubleComplex
+
+
+cdef extern from * nogil:
+    """
+    template<typename T>
+    class nullable_unique_ptr {
+      public:
+        nullable_unique_ptr() noexcept = default;
+
+        nullable_unique_ptr(std::nullptr_t) noexcept = delete;
+
+        explicit nullable_unique_ptr(T* data, bool own_data):
+            own_data_(own_data)
+        {
+            if (own_data)
+                manager_.reset(data);
+            else
+                raw_data_ = data;
+        }
+
+        nullable_unique_ptr(const nullable_unique_ptr&) = delete;
+
+        nullable_unique_ptr& operator=(const nullable_unique_ptr&) = delete;
+
+        nullable_unique_ptr(nullable_unique_ptr&& other) noexcept
+        {
+            own_data_ = other.own_data_;
+            other.own_data_ = false;  // ownership is transferred
+            if (own_data_)
+            {
+                manager_ = std::move(other.manager_);
+                raw_data_ = nullptr;  // just in case
+            }
+            else
+            {
+                manager_.reset(nullptr);  // just in case
+                raw_data_ = other.raw_data_;
+            }
+        }
+
+        nullable_unique_ptr& operator=(nullable_unique_ptr&& other) noexcept
+        {
+            own_data_ = other.own_data_;
+            other.own_data_ = false;  // ownership is transferred
+            if (own_data_)
+            {
+                manager_ = std::move(other.manager_);
+                raw_data_ = nullptr;  // just in case
+            }
+            else
+            {
+                manager_.reset(nullptr);  // just in case
+                raw_data_ = other.raw_data_;
+            }
+            return *this;
+        }
+
+        ~nullable_unique_ptr() = default;
+
+        void reset(T* data, bool own_data)
+        {
+            own_data_ = own_data;
+            if (own_data_)
+            {
+                manager_.reset(data);
+                raw_data_ = nullptr;
+            }
+            else
+            {
+                manager_.reset(nullptr);
+                raw_data_ = data;
+            }
+        }
+
+        void swap(nullable_unique_ptr& other) noexcept
+        {
+            std::swap(manager_, other.manager_);
+            std::swap(raw_data_, other.raw_data_);
+            std::swap(own_data_, other.own_data_);
+        }
+
+        /*
+         * Get the pointer to the underlying object (this is different from data()!).
+         */
+        T* get() const noexcept
+        {
+            if (own_data_)
+                return manager_.get();
+            else
+                return raw_data_;
+        }
+
+        /*
+         * Get the pointer to the underlying buffer (this is different from get()!).
+         */
+        void* data() noexcept
+        {
+            if (own_data_)
+                return manager_.get()->data();
+            else
+                return raw_data_;
+        }
+
+        T& operator*()
+        {
+            if (own_data_)
+                return *manager_;
+            else
+                return *raw_data_;
+        }
+
+      private:
+        std::unique_ptr<T> manager_{};
+        T* raw_data_{nullptr};
+        bool own_data_{false};
+    };
+    """
+    # xref: cython/Cython/Includes/libcpp/memory.pxd
+    cdef cppclass nullable_unique_ptr[T]:
+        nullable_unique_ptr()
+        nullable_unique_ptr(T*, cppbool)
+        nullable_unique_ptr(nullable_unique_ptr[T]&)
+
+        # Modifiers
+        void reset(T*, cppbool)
+        void swap(nullable_unique_ptr&)
+
+        # Observers
+        T* get()
+        T& operator*()
+        void* data()
+
+
+ctypedef fused ResT:
+    intptr_t
+    int
+    int32_t
+    int64_t
+    uint32_t
+    float
+    double
+    size_t
+
+
+ctypedef fused PtrT:
+    float
+    double
+    cuComplex
+    cuDoubleComplex
+    void
+
+
+cdef cppclass nested_resource[T]:
+    nullable_unique_ptr[ vector[intptr_t] ] ptrs
+    nullable_unique_ptr[ vector[vector[T]] ] nested_resource_ptr
+
+
+# accepts the output pointer as input to use the return value for exception propagation
+cdef int get_resource_ptr(nullable_unique_ptr[vector[ResT]] &in_out_ptr, object obj, ResT* __unused) except 1
+cdef int get_resource_ptrs(nullable_unique_ptr[ vector[PtrT*] ] &in_out_ptr, object obj, PtrT* __unused) except 1
+cdef int get_nested_resource_ptr(nested_resource[ResT] &in_out_ptr, object obj, ResT* __unused) except 1
+
+cdef bint is_nested_sequence(data)
+cdef void* get_buffer_pointer(buf, Py_ssize_t size, readonly=*) except*

nvmath/bindings/_internal/utils.pyi

@@ -0,0 +1,10 @@
+# Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES. ALL RIGHTS RESERVED.
+#
+# SPDX-License-Identifier: Apache-2.0
+
+__pyx_capi__: dict
+__test__: dict
+
+class FunctionNotFoundError(RuntimeError): ...
+
+class NotSupportedError(RuntimeError): ...