| /libCEED/backends/cuda-ref/ |
| H A D | ceed-cuda-ref-basis.c | 147 if (data->d_points_per_elem) CeedCallCuda(ceed, cudaFree(data->d_points_per_elem)); in CeedBasisApplyAtPointsCore_Cuda() 148 CeedCallCuda(ceed, cudaMalloc((void **)&data->d_points_per_elem, num_bytes)); in CeedBasisApplyAtPointsCore_Cuda() 154 …CeedCallCuda(ceed, cudaMemcpy(data->d_points_per_elem, num_points, num_bytes, cudaMemcpyHostToDevi… in CeedBasisApplyAtPointsCore_Cuda() 173 CeedCallCuda(ceed, cudaMalloc((void **)&data->d_chebyshev_interp_1d, interp_bytes)); in CeedBasisApplyAtPointsCore_Cuda() 174 …CeedCallCuda(ceed, cudaMemcpy(data->d_chebyshev_interp_1d, chebyshev_interp_1d, interp_bytes, cuda… in CeedBasisApplyAtPointsCore_Cuda() 182 if (data->moduleAtPoints) CeedCallCuda(ceed, cuModuleUnload(data->moduleAtPoints)); in CeedBasisApplyAtPointsCore_Cuda() 363 CeedCallCuda(ceed, cuModuleUnload(data->module)); in CeedBasisDestroy_Cuda() 364 if (data->moduleAtPoints) CeedCallCuda(ceed, cuModuleUnload(data->moduleAtPoints)); in CeedBasisDestroy_Cuda() 365 if (data->d_q_weight_1d) CeedCallCuda(ceed, cudaFree(data->d_q_weight_1d)); in CeedBasisDestroy_Cuda() 367 if (data->d_points_per_elem) CeedCallCuda(ceed, cudaFree(data->d_points_per_elem)); in CeedBasisDestroy_Cuda() [all …]
|
| H A D | ceed-cuda-ref-restriction.c | 393 CeedCallCuda(ceed, cuModuleUnload(impl->module)); in CeedElemRestrictionDestroy_Cuda() 396 CeedCallCuda(ceed, cudaFree((CeedInt *)impl->d_offsets_owned)); in CeedElemRestrictionDestroy_Cuda() 397 CeedCallCuda(ceed, cudaFree((CeedInt *)impl->d_t_offsets)); in CeedElemRestrictionDestroy_Cuda() 398 CeedCallCuda(ceed, cudaFree((CeedInt *)impl->d_t_indices)); in CeedElemRestrictionDestroy_Cuda() 399 CeedCallCuda(ceed, cudaFree((CeedInt *)impl->d_l_vec_indices)); in CeedElemRestrictionDestroy_Cuda() 401 CeedCallCuda(ceed, cudaFree((bool *)impl->d_orients_owned)); in CeedElemRestrictionDestroy_Cuda() 403 CeedCallCuda(ceed, cudaFree((CeedInt8 *)impl->d_curl_orients_owned)); in CeedElemRestrictionDestroy_Cuda() 405 CeedCallCuda(ceed, cudaFree((CeedInt *)impl->d_offsets_at_points_owned)); in CeedElemRestrictionDestroy_Cuda() 407 CeedCallCuda(ceed, cudaFree((CeedInt *)impl->d_points_per_elem_owned)); in CeedElemRestrictionDestroy_Cuda() 477 CeedCallCuda(ceed, cudaMalloc((void **)&impl->d_l_vec_indices, num_nodes * sizeof(CeedInt))); in CeedElemRestrictionOffset_Cuda() [all …]
|
| H A D | ceed-cuda-ref-qfunctioncontext.c | 36 CeedCallCuda(ceed, cudaMalloc((void **)&impl->d_data_owned, ctx_size)); in CeedQFunctionContextSyncH2D_Cuda() 39 CeedCallCuda(ceed, cudaMemcpy(impl->d_data, impl->h_data, ctx_size, cudaMemcpyHostToDevice)); in CeedQFunctionContextSyncH2D_Cuda() 67 CeedCallCuda(ceed, cudaMemcpy(impl->h_data, impl->d_data, ctx_size, cudaMemcpyDeviceToHost)); in CeedQFunctionContextSyncD2H_Cuda() 188 CeedCallCuda(ceed, cudaFree(impl->d_data_owned)); in CeedQFunctionContextSetDataDevice_Cuda() 194 CeedCallCuda(ceed, cudaMalloc((void **)&impl->d_data_owned, ctx_size)); in CeedQFunctionContextSetDataDevice_Cuda() 197 CeedCallCuda(ceed, cudaMemcpy(impl->d_data, data, ctx_size, cudaMemcpyDeviceToDevice)); in CeedQFunctionContextSetDataDevice_Cuda() 320 CeedCallCuda(CeedQFunctionContextReturnCeed(ctx), cudaFree(impl->d_data_owned)); in CeedQFunctionContextDestroy_Cuda()
|
| H A D | ceed-cuda-ref-operator.c | 65 CeedCallCuda(ceed, cuModuleUnload(impl->diag->module)); in CeedOperatorDestroy_Cuda() 68 CeedCallCuda(ceed, cuModuleUnload(impl->diag->module_point_block)); in CeedOperatorDestroy_Cuda() 70 CeedCallCuda(ceed, cudaFree(impl->diag->d_eval_modes_in)); in CeedOperatorDestroy_Cuda() 71 CeedCallCuda(ceed, cudaFree(impl->diag->d_eval_modes_out)); in CeedOperatorDestroy_Cuda() 72 CeedCallCuda(ceed, cudaFree(impl->diag->d_identity)); in CeedOperatorDestroy_Cuda() 73 CeedCallCuda(ceed, cudaFree(impl->diag->d_interp_in)); in CeedOperatorDestroy_Cuda() 74 CeedCallCuda(ceed, cudaFree(impl->diag->d_interp_out)); in CeedOperatorDestroy_Cuda() 75 CeedCallCuda(ceed, cudaFree(impl->diag->d_grad_in)); in CeedOperatorDestroy_Cuda() 76 CeedCallCuda(ceed, cudaFree(impl->diag->d_grad_out)); in CeedOperatorDestroy_Cuda() 77 CeedCallCuda(ceed, cudaFree(impl->diag->d_div_in)); in CeedOperatorDestroy_Cuda() [all …]
|
| H A D | ceed-cuda-ref-vector.c | 57 CeedCallCuda(CeedVectorReturnCeed(vec), cudaMalloc((void **)&impl->d_array_owned, bytes)); in CeedVectorSyncH2D_Cuda() 60 …CeedCallCuda(CeedVectorReturnCeed(vec), cudaMemcpy(impl->d_array, impl->h_array, bytes, cudaMemcpy… in CeedVectorSyncH2D_Cuda() 90 …CeedCallCuda(CeedVectorReturnCeed(vec), cudaMemcpy(impl->h_array, impl->d_array, bytes, cudaMemcpy… in CeedVectorSyncD2H_Cuda() 325 …CeedCallCuda(CeedVectorReturnCeed(vec), cudaMemset(impl->d_array, 0, length * sizeof(CeedScalar))); in CeedVectorSetValue_Cuda() 585 …CeedCallCuda(ceed, cudaMemcpy(&norm_no_abs, impl->d_array + index - 1, sizeof(CeedScalar), cudaMem… in CeedVectorNorm_Cuda() 598 …CeedCallCuda(ceed, cudaMemcpy(&sub_max, d_array_start + index - 1, sizeof(CeedScalar), cudaMemcpyD… in CeedVectorNorm_Cuda() 609 …CeedCallCuda(ceed, cudaMemcpy(&norm_no_abs, impl->d_array + index - 1, sizeof(CeedScalar), cudaMem… in CeedVectorNorm_Cuda() 622 …CeedCallCuda(ceed, cudaMemcpy(&sub_max, d_array_start + index - 1, sizeof(CeedScalar), cudaMemcpyD… in CeedVectorNorm_Cuda() 840 CeedCallCuda(CeedVectorReturnCeed(vec), cudaFree(impl->d_array_owned)); in CeedVectorDestroy_Cuda()
|
| H A D | ceed-cuda-ref-qfunction.c | 72 if (data->module) CeedCallCuda(CeedQFunctionReturnCeed(qf), cuModuleUnload(data->module)); in CeedQFunctionDestroy_Cuda()
|
| /libCEED/backends/cuda/ |
| H A D | ceed-cuda-common.c | 25 CeedCallCuda(ceed, cudaGetDevice(¤t_device_id)); in CeedInit_Cuda() 27 CeedCallCuda(ceed, cudaSetDevice(device_id)); in CeedInit_Cuda() 33 CeedCallCuda(ceed, cudaGetDeviceProperties(&data->device_prop, current_device_id)); in CeedInit_Cuda() 60 …if (!*(void **)target_array_owned) CeedCallCuda(ceed, cudaMalloc(target_array_owned, size_unit * n… in CeedSetDeviceGenericArray_Cuda() 64 …if (source_array) CeedCallCuda(ceed, cudaMemcpy(*(void **)target_array, source_array, size_unit * … in CeedSetDeviceGenericArray_Cuda() 67 CeedCallCuda(ceed, cudaFree(*(void **)target_array_owned)); in CeedSetDeviceGenericArray_Cuda() 73 CeedCallCuda(ceed, cudaFree(*(void **)target_array_owned)); in CeedSetDeviceGenericArray_Cuda()
|
| H A D | ceed-cuda-compile.cpp | 113 CeedCallCuda(ceed, cudaGetDeviceProperties(&prop, ceed_data->device_id)); in CeedCompileCore_Cuda() 221 CeedCallCuda(ceed, cuModuleLoadData(module, ptx)); in CeedCompileCore_Cuda() 434 CeedCallCuda(ceed, cuModuleGetFunction(kernel, module, name)); in CeedGetKernel_Cuda() 446 …CeedCallCuda(ceed, cuOccupancyMaxPotentialBlockSize(&min_grid_size, &max_block_size, kernel, NULL,… in CeedRunKernelAutoblockCuda()
|
| H A D | ceed-cuda-common.h | 35 #define CeedCallCuda(ceed, ...) \ macro
|
| /libCEED/backends/cuda-shared/ |
| H A D | ceed-cuda-shared-basis.c | 255 if (data->d_points_per_elem) CeedCallCuda(ceed, cudaFree(data->d_points_per_elem)); in CeedBasisApplyAtPointsCore_Cuda_shared() 256 CeedCallCuda(ceed, cudaMalloc((void **)&data->d_points_per_elem, num_bytes)); in CeedBasisApplyAtPointsCore_Cuda_shared() 262 …CeedCallCuda(ceed, cudaMemcpy(data->d_points_per_elem, num_points, num_bytes, cudaMemcpyHostToDevi… in CeedBasisApplyAtPointsCore_Cuda_shared() 281 CeedCallCuda(ceed, cudaMalloc((void **)&data->d_chebyshev_interp_1d, interp_bytes)); in CeedBasisApplyAtPointsCore_Cuda_shared() 282 …CeedCallCuda(ceed, cudaMemcpy(data->d_chebyshev_interp_1d, chebyshev_interp_1d, interp_bytes, cuda… in CeedBasisApplyAtPointsCore_Cuda_shared() 290 if (data->moduleAtPoints) CeedCallCuda(ceed, cuModuleUnload(data->moduleAtPoints)); in CeedBasisApplyAtPointsCore_Cuda_shared() 577 CeedCallCuda(ceed, cuModuleUnload(data->module)); in CeedBasisDestroy_Cuda_shared() 578 if (data->moduleAtPoints) CeedCallCuda(ceed, cuModuleUnload(data->moduleAtPoints)); in CeedBasisDestroy_Cuda_shared() 579 if (data->d_q_weight_1d) CeedCallCuda(ceed, cudaFree(data->d_q_weight_1d)); in CeedBasisDestroy_Cuda_shared() 581 if (data->d_points_per_elem) CeedCallCuda(ceed, cudaFree(data->d_points_per_elem)); in CeedBasisDestroy_Cuda_shared() [all …]
|
| /libCEED/backends/cuda-gen/ |
| H A D | ceed-cuda-gen-operator.c | 30 if (impl->module) CeedCallCuda(ceed, cuModuleUnload(impl->module)); in CeedOperatorDestroy_Cuda_gen() 31 if (impl->module_assemble_full) CeedCallCuda(ceed, cuModuleUnload(impl->module_assemble_full)); in CeedOperatorDestroy_Cuda_gen() 32 …if (impl->module_assemble_diagonal) CeedCallCuda(ceed, cuModuleUnload(impl->module_assemble_diagon… in CeedOperatorDestroy_Cuda_gen() 33 …if (impl->module_assemble_qfunction) CeedCallCuda(ceed, cuModuleUnload(impl->module_assemble_qfunc… in CeedOperatorDestroy_Cuda_gen() 34 if (impl->points.num_per_elem) CeedCallCuda(ceed, cudaFree((void **)impl->points.num_per_elem)); in CeedOperatorDestroy_Cuda_gen() 192 … if (data->points.num_per_elem) CeedCallCuda(ceed, cudaFree((void **)data->points.num_per_elem)); in CeedOperatorApplyAddCore_Cuda_gen() 193 CeedCallCuda(ceed, cudaMalloc((void **)&data->points.num_per_elem, num_bytes)); in CeedOperatorApplyAddCore_Cuda_gen() 194 …CeedCallCuda(ceed, cudaMemcpy((void *)data->points.num_per_elem, points_per_elem, num_bytes, cudaM… in CeedOperatorApplyAddCore_Cuda_gen() 208 …CeedCallCuda(ceed, cuOccupancyMaxPotentialBlockSize(&min_grid_size, &max_threads_per_block, data->… in CeedOperatorApplyAddCore_Cuda_gen() 311 if (is_sequential) CeedCallCuda(ceed, cudaStreamCreate(&stream)); in CeedOperatorApplyAddComposite_Cuda_gen() [all …]
|
| H A D | ceed-cuda-gen-qfunction.c | 29 CeedCallCuda(CeedQFunctionReturnCeed(qf), cudaFree(data->d_c)); in CeedQFunctionDestroy_Cuda_gen()
|
| H A D | ceed-cuda-gen-operator-build.cpp | 255 …CeedCallCuda(CeedBasisReturnCeed(basis), cudaMalloc((void **)&basis_data->d_chebyshev_interp_1d, i… in CeedOperatorBuildKernelFieldData_Cuda_gen() 256 CeedCallCuda(CeedBasisReturnCeed(basis), in CeedOperatorBuildKernelFieldData_Cuda_gen() 293 …CeedCallCuda(CeedBasisReturnCeed(basis), cudaMalloc((void **)&basis_data->d_chebyshev_interp_1d, i… in CeedOperatorBuildKernelFieldData_Cuda_gen() 294 CeedCallCuda(CeedBasisReturnCeed(basis), in CeedOperatorBuildKernelFieldData_Cuda_gen()
|
| /libCEED/backends/magma/ |
| H A D | ceed-magma-basis.c | 490 CeedCallCuda(ceed, cuModuleUnload(impl->module)); in CeedBasisDestroy_Magma() 514 CeedCallCuda(ceed, cuModuleUnload(impl->module[in])); in CeedBasisDestroyNonTensor_Magma()
|