From: Bruno Turcksin Date: Fri, 22 Jun 2018 12:56:11 +0000 (-0400) Subject: Move functions in cuda_solver_direc from internal to anonymous namespace X-Git-Tag: v9.1.0-rc1~1000^2 X-Git-Url: https://gitweb.dealii.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=00ee4a1ba8eeced5713985684f22469d520862b9;p=dealii.git Move functions in cuda_solver_direc from internal to anonymous namespace --- diff --git a/include/deal.II/lac/cuda_solver_direct.h b/include/deal.II/lac/cuda_solver_direct.h index 86acf16fc2..2a4956918c 100644 --- a/include/deal.II/lac/cuda_solver_direct.h +++ b/include/deal.II/lac/cuda_solver_direct.h @@ -100,9 +100,9 @@ namespace CUDAWrappers /** * Reference to the object that controls convergence of the iterative - * solver. In fact, for these Trilinos wrappers, Trilinos does so itself, - * but we copy the data from this object before starting the solution - * process, and copy the data back into it afterwards. + * solver. In fact, for these CUDA wrappers, cuSOLVER and cuSPARSE do so + * themselve, but we copy the data from this object before starting the + * solution process, and copy the data back into it afterwards. */ SolverControl &solver_control; diff --git a/source/lac/cuda_solver_direct.cu b/source/lac/cuda_solver_direct.cu index dfdf9bb647..f7a8ee233a 100644 --- a/source/lac/cuda_solver_direct.cu +++ b/source/lac/cuda_solver_direct.cu @@ -19,7 +19,7 @@ DEAL_II_NAMESPACE_OPEN namespace CUDAWrappers { - namespace internal + namespace { void cusparsecsr2dense(cusparseHandle_t cusparse_handle, @@ -28,7 +28,7 @@ namespace CUDAWrappers { auto cusparse_matrix = matrix.get_cusparse_matrix(); - cusparseStatus_t cusparse_error_code = + const cusparseStatus_t cusparse_error_code = cusparseScsr2dense(cusparse_handle, matrix.m(), matrix.n(), @@ -50,7 +50,7 @@ namespace CUDAWrappers { auto cusparse_matrix = matrix.get_cusparse_matrix(); - cusparseStatus_t cusparse_error_code = + const cusparseStatus_t cusparse_error_code = cusparseDcsr2dense(cusparse_handle, matrix.m(), matrix.n(), @@ -72,7 +72,7 @@ namespace CUDAWrappers float * dense_matrix_dev, int & workspace_size) { - cusolverStatus_t cusolver_error_code = cusolverDnSgetrf_bufferSize( + const cusolverStatus_t cusolver_error_code = cusolverDnSgetrf_bufferSize( cusolver_dn_handle, m, n, dense_matrix_dev, m, &workspace_size); AssertCusolver(cusolver_error_code); } @@ -86,7 +86,7 @@ namespace CUDAWrappers double * dense_matrix_dev, int & workspace_size) { - cusolverStatus_t cusolver_error_code = cusolverDnDgetrf_bufferSize( + const cusolverStatus_t cusolver_error_code = cusolverDnDgetrf_bufferSize( cusolver_dn_handle, m, n, dense_matrix_dev, m, &workspace_size); AssertCusolver(cusolver_error_code); } @@ -102,7 +102,7 @@ namespace CUDAWrappers int * pivot_dev, int * info_dev) { - cusolverStatus_t cusolver_error_code = + const cusolverStatus_t cusolver_error_code = cusolverDnSgetrf(cusolver_dn_handle, m, n, @@ -125,7 +125,7 @@ namespace CUDAWrappers int * pivot_dev, int * info_dev) { - cusolverStatus_t cusolver_error_code = + const cusolverStatus_t cusolver_error_code = cusolverDnDgetrf(cusolver_dn_handle, m, n, @@ -147,8 +147,8 @@ namespace CUDAWrappers float * b, int * info_dev) { - const int n_rhs = 1; - cusolverStatus_t cusolver_error_code = + const int n_rhs = 1; + const cusolverStatus_t cusolver_error_code = cusolverDnSgetrs(cusolver_dn_handle, CUBLAS_OP_N, m, @@ -172,8 +172,8 @@ namespace CUDAWrappers double * b, int * info_dev) { - const int n_rhs = 1; - cusolverStatus_t cusolver_error_code = + const int n_rhs = 1; + const cusolverStatus_t cusolver_error_code = cusolverDnDgetrs(cusolver_dn_handle, CUBLAS_OP_N, m, @@ -200,8 +200,8 @@ namespace CUDAWrappers const float * b_host, float * x_host) { - int singularity = 0; - cusolverStatus_t cusolver_error_code = + int singularity = 0; + const cusolverStatus_t cusolver_error_code = cusolverSpScsrlsvluHost(cusolver_sp_handle, n_rows, nnz, @@ -231,8 +231,8 @@ namespace CUDAWrappers const double * b_host, double * x_host) { - int singularity = 0; - cusolverStatus_t cusolver_error_code = + int singularity = 0; + const cusolverStatus_t cusolver_error_code = cusolverSpDcsrlsvluHost(cusolver_sp_handle, n_rows, nnz, @@ -260,7 +260,7 @@ namespace CUDAWrappers auto cusparse_matrix = matrix.get_cusparse_matrix(); int singularity = 0; - cusolverStatus_t cusolver_error_code = + const cusolverStatus_t cusolver_error_code = cusolverSpScsrlsvchol(cusolver_sp_handle, matrix.m(), matrix.n_nonzero_elements(), @@ -288,7 +288,7 @@ namespace CUDAWrappers auto cusparse_matrix = matrix.get_cusparse_matrix(); int singularity = 0; - cusolverStatus_t cusolver_error_code = + const cusolverStatus_t cusolver_error_code = cusolverSpDcsrlsvchol(cusolver_sp_handle, matrix.m(), matrix.n_nonzero_elements(), @@ -323,11 +323,11 @@ namespace CUDAWrappers Utilities::CUDA::malloc(dense_matrix_dev, m * n); // Change the format of matrix to dense - internal::cusparsecsr2dense(cusparse_handle, matrix, dense_matrix_dev); + cusparsecsr2dense(cusparse_handle, matrix, dense_matrix_dev); // Create the working space int workspace_size = 0; - internal::cusolverDngetrf_buffer_size( + cusolverDngetrf_buffer_size( cusolver_dn_handle, m, n, dense_matrix_dev, workspace_size); Assert(workspace_size > 0, ExcMessage("No workspace was allocated")); Number *workspace_dev; @@ -339,13 +339,13 @@ namespace CUDAWrappers int *info_dev; Utilities::CUDA::malloc(info_dev, 1); - internal::cusolverDngetrf(cusolver_dn_handle, - m, - n, - dense_matrix_dev, - workspace_dev, - pivot_dev, - info_dev); + cusolverDngetrf(cusolver_dn_handle, + m, + n, + dense_matrix_dev, + workspace_dev, + pivot_dev, + info_dev); #ifdef DEBUG int info = 0; @@ -360,7 +360,7 @@ namespace CUDAWrappers cudaError_t cuda_error_code = cudaMemcpy(x_dev, b_dev, m * sizeof(Number), cudaMemcpyDeviceToDevice); AssertCuda(cuda_error_code); - internal::cusolverDngetrs( + cusolverDngetrs( cusolver_dn_handle, m, dense_matrix_dev, pivot_dev, x_dev, info_dev); #ifdef DEBUG cuda_error_code = @@ -403,20 +403,20 @@ namespace CUDAWrappers std::vector x_host(n_rows); Utilities::CUDA::copy_to_host(x_dev, x_host); - internal::cusolverSpcsrlsvluHost(cusolver_sp_handle, - n_rows, - nnz, - std::get<3>(cusparse_matrix), - val_host.data(), - row_ptr_host.data(), - column_index_host.data(), - b_host.data(), - x_host.data()); + cusolverSpcsrlsvluHost(cusolver_sp_handle, + n_rows, + nnz, + std::get<3>(cusparse_matrix), + val_host.data(), + row_ptr_host.data(), + column_index_host.data(), + b_host.data(), + x_host.data()); // Move the solution back to the device Utilities::CUDA::copy_to_dev(x_host, x_dev); } - } // namespace internal + } // namespace @@ -456,21 +456,21 @@ namespace CUDAWrappers const LinearAlgebra::CUDAWrappers::Vector &b) { if (additional_data.solver_type == "Cholesky") - internal::cholesky_factorization(cuda_handle.cusolver_sp_handle, - A, - b.get_values(), - x.get_values()); + cholesky_factorization(cuda_handle.cusolver_sp_handle, + A, + b.get_values(), + x.get_values()); else if (additional_data.solver_type == "LU_dense") - internal::lu_factorization(cuda_handle.cusparse_handle, - cuda_handle.cusolver_dn_handle, - A, - b.get_values(), - x.get_values()); + lu_factorization(cuda_handle.cusparse_handle, + cuda_handle.cusolver_dn_handle, + A, + b.get_values(), + x.get_values()); else if (additional_data.solver_type == "LU_host") - internal::lu_factorization(cuda_handle.cusolver_sp_handle, - A, - b.get_values(), - x.get_values()); + lu_factorization(cuda_handle.cusolver_sp_handle, + A, + b.get_values(), + x.get_values()); else AssertThrow(false, ExcMessage("The provided solver name " +