/* ************************************************************************ * Copyright (c) 2020 Advanced Micro Devices, Inc. * * Permission is hereby granted, free of charge, to any person obtaining a copy * of this software and associated documentation files (the "Software"), to deal * in the Software without restriction, including without limitation the rights * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell * copies of the Software, and to permit persons to whom the Software is * furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN * THE SOFTWARE. * * ************************************************************************ */ #pragma once #ifndef TESTING_DENSE_TO_SPARSE_CSC_HPP #define TESTING_DENSE_TO_SPARSE_CSC_HPP #include "hipsparse_test_unique_ptr.hpp" #include "unit.hpp" #include "utility.hpp" #include #include #include using namespace hipsparse_test; void testing_dense_to_sparse_csc_bad_arg(void) { #if(!defined(CUDART_VERSION)) int64_t safe_size = 100; int32_t m = 10; int32_t n = 10; int64_t nnz = 10; int64_t ld = m; hipsparseIndexBase_t idxBase = HIPSPARSE_INDEX_BASE_ZERO; hipsparseDenseToSparseAlg_t alg = HIPSPARSE_DENSETOSPARSE_ALG_DEFAULT; hipsparseOrder_t order = HIPSPARSE_ORDER_COLUMN; // Index and data type hipsparseIndexType_t iType = HIPSPARSE_INDEX_64I; hipsparseIndexType_t jType = HIPSPARSE_INDEX_32I; hipDataType dataType = HIP_R_32F; // Create handle std::unique_ptr unique_ptr_handle(new handle_struct); hipsparseHandle_t handle = unique_ptr_handle->handle; auto ddense_val_managed = hipsparse_unique_ptr{device_malloc(sizeof(float) * safe_size), device_free}; auto dcsc_col_ptr_managed = hipsparse_unique_ptr{device_malloc(sizeof(int64_t) * safe_size), device_free}; auto dcsc_row_ind_managed = hipsparse_unique_ptr{device_malloc(sizeof(int32_t) * safe_size), device_free}; auto dcsc_val_managed = hipsparse_unique_ptr{device_malloc(sizeof(float) * safe_size), device_free}; auto dbuf_managed = hipsparse_unique_ptr{device_malloc(sizeof(char) * safe_size), device_free}; float* ddense_val = (float*)ddense_val_managed.get(); int64_t* dcsc_col_ptr = (int64_t*)dcsc_col_ptr_managed.get(); int32_t* dcsc_row_ind = (int32_t*)dcsc_row_ind_managed.get(); float* dcsc_val = (float*)dcsc_val_managed.get(); void* dbuf = (void*)dbuf_managed.get(); if(!ddense_val || !dcsc_col_ptr || !dcsc_row_ind || !dcsc_val || !dbuf) { PRINT_IF_HIP_ERROR(hipErrorOutOfMemory); return; } // Matrix structures hipsparseDnVecDescr_t matA; hipsparseSpMatDescr_t matB; size_t bsize; // Create matrix structures verify_hipsparse_status_success( hipsparseCreateDnMat(&matA, m, n, ld, ddense_val, dataType, order), "success"); verify_hipsparse_status_success(hipsparseCreateCsc(&matB, m, n, nnz, dcsc_col_ptr, dcsc_row_ind, dcsc_val, iType, jType, idxBase, dataType), "success"); // denseToSparse buffer size verify_hipsparse_status_invalid_handle( hipsparseDenseToSparse_bufferSize(nullptr, matA, matB, alg, &bsize)); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_bufferSize(handle, nullptr, matB, alg, &bsize), "Error: matA is nullptr"); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_bufferSize(handle, matA, nullptr, alg, &bsize), "Error: matB is nullptr"); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_bufferSize(handle, matA, matB, alg, nullptr), "Error: bsize is nullptr"); // denseToSparse analysis verify_hipsparse_status_invalid_handle( hipsparseDenseToSparse_analysis(nullptr, matA, matB, alg, &bsize)); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_analysis(handle, nullptr, matB, alg, &bsize), "Error: matA is nullptr"); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_analysis(handle, matA, nullptr, alg, &bsize), "Error: matB is nullptr"); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_analysis(handle, matA, matB, alg, nullptr), "Error: bsize is nullptr"); // denseToSparse_convert verify_hipsparse_status_invalid_handle( hipsparseDenseToSparse_convert(nullptr, matA, matB, alg, dbuf)); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_convert(handle, nullptr, matB, alg, dbuf), "Error: matA is nullptr"); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_convert(handle, matA, nullptr, alg, dbuf), "Error: matB is nullptr"); verify_hipsparse_status_invalid_pointer( hipsparseDenseToSparse_convert(handle, matA, matB, alg, nullptr), "Error: dbuf is nullptr"); // Destruct verify_hipsparse_status_success(hipsparseDestroyDnMat(matA), "success"); verify_hipsparse_status_success(hipsparseDestroySpMat(matB), "success"); #endif } template hipsparseStatus_t testing_dense_to_sparse_csc(void) { #if(!defined(CUDART_VERSION)) hipsparseIndexBase_t idx_base = HIPSPARSE_INDEX_BASE_ZERO; hipsparseDenseToSparseAlg_t alg = HIPSPARSE_DENSETOSPARSE_ALG_DEFAULT; hipsparseOrder_t order = HIPSPARSE_ORDER_COLUMN; // Index and data type hipsparseIndexType_t typeI = (typeid(I) == typeid(int32_t)) ? HIPSPARSE_INDEX_32I : HIPSPARSE_INDEX_64I; hipsparseIndexType_t typeJ = (typeid(J) == typeid(int32_t)) ? HIPSPARSE_INDEX_32I : HIPSPARSE_INDEX_64I; hipDataType typeT = (typeid(T) == typeid(float)) ? HIP_R_32F : ((typeid(T) == typeid(double)) ? HIP_R_64F : ((typeid(T) == typeid(hipComplex) ? HIP_C_32F : HIP_C_64F))); // hipSPARSE handle std::unique_ptr test_handle(new handle_struct); hipsparseHandle_t handle = test_handle->handle; J m = 100; J n = 100; I ld = m; // Host structures std::vector hdense_val(ld * n); for(int i = 0; i < ld; ++i) { for(int j = 0; j < n; ++j) { hdense_val[j * ld + i] = make_DataType(-1); } } srand(0); gen_dense_random_sparsity_pattern(m, n, hdense_val.data(), ld, 0.2); // allocate memory on device auto dptr_managed = hipsparse_unique_ptr{device_malloc(sizeof(I) * (n + 1)), device_free}; auto ddense_managed = hipsparse_unique_ptr{device_malloc(sizeof(T) * ld * n), device_free}; I* dptr = (I*)dptr_managed.get(); T* ddense = (T*)ddense_managed.get(); if(!dptr || !ddense) { verify_hipsparse_status_success(HIPSPARSE_STATUS_ALLOC_FAILED, "!dptr || !ddense"); return HIPSPARSE_STATUS_ALLOC_FAILED; } // Copy host dense matrix to device CHECK_HIP_ERROR( hipMemcpy(ddense, hdense_val.data(), sizeof(T) * ld * n, hipMemcpyHostToDevice)); // Create dense matrix hipsparseDnMatDescr_t matA; CHECK_HIPSPARSE_ERROR(hipsparseCreateDnMat(&matA, m, n, ld, ddense, typeT, order)); // Create matrices hipsparseSpMatDescr_t matB; CHECK_HIPSPARSE_ERROR( hipsparseCreateCsc(&matB, m, n, 0, dptr, nullptr, nullptr, typeI, typeJ, idx_base, typeT)); // Query DenseToSparse buffer size_t bufferSize; CHECK_HIPSPARSE_ERROR(hipsparseDenseToSparse_bufferSize(handle, matA, matB, alg, &bufferSize)); void* buffer; CHECK_HIP_ERROR(hipMalloc(&buffer, bufferSize)); CHECK_HIPSPARSE_ERROR(hipsparseDenseToSparse_analysis(handle, matA, matB, alg, buffer)); int64_t rows, cols, nnz; CHECK_HIPSPARSE_ERROR(hipsparseSpMatGetSize(matB, &rows, &cols, &nnz)); auto drow_managed = hipsparse_unique_ptr{device_malloc(sizeof(J) * nnz), device_free}; auto dval_managed = hipsparse_unique_ptr{device_malloc(sizeof(T) * nnz), device_free}; J* drow = (J*)drow_managed.get(); T* dval = (T*)dval_managed.get(); if(!drow || !dval) { verify_hipsparse_status_success(HIPSPARSE_STATUS_ALLOC_FAILED, "!drow || !dval"); return HIPSPARSE_STATUS_ALLOC_FAILED; } CHECK_HIPSPARSE_ERROR(hipsparseCscSetPointers(matB, dptr, drow, dval)); CHECK_HIPSPARSE_ERROR(hipsparseDenseToSparse_convert(handle, matA, matB, alg, buffer)); // copy output from device to CPU std::vector hcsc_col_ptr(n + 1); std::vector hcsc_row_ind(nnz); std::vector hcsc_val(nnz); CHECK_HIP_ERROR( hipMemcpy(hcsc_col_ptr.data(), dptr, sizeof(I) * (n + 1), hipMemcpyDeviceToHost)); CHECK_HIP_ERROR(hipMemcpy(hcsc_row_ind.data(), drow, sizeof(J) * nnz, hipMemcpyDeviceToHost)); CHECK_HIP_ERROR(hipMemcpy(hcsc_val.data(), dval, sizeof(T) * nnz, hipMemcpyDeviceToHost)); // Query for warpSize hipDeviceProp_t prop; hipGetDeviceProperties(&prop, 0); std::vector hcsc_col_ptr_cpu(n + 1); std::vector hcsc_row_ind_cpu(nnz); std::vector hcsc_val_cpu(nnz); std::vector hnnz_per_column(n, 0); for(J i = 0; i < m; ++i) { for(J j = 0; j < n; ++j) { if(hdense_val[j * ld + i] != make_DataType(0.0)) { hnnz_per_column[j]++; } } } hcsc_col_ptr_cpu[0] = idx_base; for(J i = 0; i < n; ++i) { hcsc_col_ptr_cpu[i + 1] = hnnz_per_column[i] + hcsc_col_ptr_cpu[i]; } int index = 0; for(J i = 0; i < n; ++i) { for(J j = 0; j < m; ++j) { if(hdense_val[i * ld + j] != make_DataType(0.0)) { hcsc_val_cpu[index] = hdense_val[i * ld + j]; hcsc_row_ind_cpu[index] = j + idx_base; index++; } } } unit_check_general(1, (n + 1), 1, hcsc_col_ptr_cpu.data(), hcsc_col_ptr.data()); unit_check_general(1, nnz, 1, hcsc_row_ind_cpu.data(), hcsc_row_ind.data()); unit_check_general(1, nnz, 1, hcsc_val_cpu.data(), hcsc_val.data()); CHECK_HIP_ERROR(hipFree(buffer)); CHECK_HIPSPARSE_ERROR(hipsparseDestroyDnMat(matA)); CHECK_HIPSPARSE_ERROR(hipsparseDestroySpMat(matB)); #endif return HIPSPARSE_STATUS_SUCCESS; } #endif // TESTING_DENSE_TO_SPARSE_CSC_HPP