error LNK2019: unresolved external symbol cublasDestroy_v2

3 次查看（过去 30 天）
Zainub
Zainub 2015-2-6
评论： Joss Knight 2015-2-23
Hi.... I am using matlab 14 . when I run a cuda file (.cu) , error comes like
Error using mex
Creating library code.lib and object code.exp
code.obj : error LNK2019: unresolved external symbol cublasDestroy_v2 referenced in
function "void __cdecl gpu_blas_mmul(float const *,float const *,float *,unsigned
__int64,unsigned __int64,unsigned __int64)" (?gpu_blas_mmul@@YAXPEBM0PEAM_K22@Z)
code.obj : error LNK2019: unresolved external symbol cublasSgemm_v2 referenced in
function "void __cdecl gpu_blas_mmul(float const *,float const *,float *,unsigned
__int64,unsigned __int64,unsigned __int64)" (?gpu_blas_mmul@@YAXPEBM0PEAM_K22@Z)
code.obj : error LNK2019: unresolved external symbol cublasCreate_v2 referenced in
function "void __cdecl gpu_blas_mmul(float const *,float const *,float *,unsigned
__int64,unsigned __int64,unsigned __int64)" (?gpu_blas_mmul@@YAXPEBM0PEAM_K22@Z)
code.mexw64 : fatal error LNK1120: 3 unresolved externals
how can I get rid of it???
  3 个评论
显示 1更早的评论
Zainub
Zainub 2015-2-8
编辑：Geoff Hayes 2015-2-22
actually I want to use cuda file(.cu) file in matlab.. I use cublas library in my code...but error comes in linking of library..my code is
if true
// Low level matrix multiplication on GPU using CUDA with CURAND and CUBLAS
// C(m,n) = A(m,k) * B(k,n)
#include <iostream>
#include <cstdlib>
#include <ctime>
#include <cublas_v2.h>
#include <curand.h>
// Fill the array A(nr_rows_A, nr_cols_A) with random numbers on GPU
void GPU_fill_rand(float *A, int nr_rows_A, int nr_cols_A) {
// Create a pseudo-random number generator
curandGenerator_t prng;
curandCreateGenerator(&prng, CURAND_RNG_PSEUDO_DEFAULT);
// Set the seed for the random number generator using the system clock
curandSetPseudoRandomGeneratorSeed(prng, (unsigned long long) clock());
// Fill the array with random numbers on the device
curandGenerateUniform(prng, A, nr_rows_A * nr_cols_A);
}
// Multiply the arrays A and B on GPU and save the result in C
// C(m,n) = A(m,k) * B(k,n)
void gpu_blas_mmul(const float *A, const float *B, float *C, const int m, const int k, const int n) {
int lda=m,ldb=k,ldc=m;
const float alf = 1;
const float bet = 0;
const float *alpha = &alf;
const float *beta = &bet;
// Create a handle for CUBLAS
cublasHandle_t handle;
cublasCreate(&handle);
// Do the actual multiplication
cublasSgemm(handle, CUBLAS_OP_N, CUBLAS_OP_N, m, n, k, alpha, A, lda, B, ldb, beta, C, ldc);
// Destroy the handle
cublasDestroy(handle);
}
//Print matrix A(nr_rows_A, nr_cols_A) storage in column-major format
void print_matrix(const float *A, int nr_rows_A, int nr_cols_A) {
for(int i = 0; i < nr_rows_A; ++i){
for(int j = 0; j < nr_cols_A; ++j){
std::cout << A[j * nr_rows_A + i] << " ";
}
std::cout << std::endl;
}
std::cout << std::endl;
}
int main() {
// Allocate 3 arrays on CPU
int nr_rows_A, nr_cols_A, nr_rows_B, nr_cols_B, nr_rows_C, nr_cols_C;
// for simplicity we are going to use square arrays
nr_rows_A = nr_cols_A = nr_rows_B = nr_cols_B = nr_rows_C = nr_cols_C = 3;
float *h_A = (float *)malloc(nr_rows_A * nr_cols_A * sizeof(float));
float *h_B = (float *)malloc(nr_rows_B * nr_cols_B * sizeof(float));
float *h_C = (float *)malloc(nr_rows_C * nr_cols_C * sizeof(float));
// Allocate 3 arrays on GPU
float *d_A, *d_B, *d_C;
cudaMalloc(&d_A,nr_rows_A * nr_cols_A * sizeof(float));
cudaMalloc(&d_B,nr_rows_B * nr_cols_B * sizeof(float));
cudaMalloc(&d_C,nr_rows_C * nr_cols_C * sizeof(float));
// If you already have useful values in A and B you can copy them in GPU:
// cudaMemcpy(d_A,h_A,nr_rows_A * nr_cols_A * sizeof(float),cudaMemcpyHostToDevice);
// cudaMemcpy(d_B,h_B,nr_rows_B * nr_cols_B * sizeof(float),cudaMemcpyHostToDevice);
// Fill the arrays A and B on GPU with random numbers
GPU_fill_rand(d_A, nr_rows_A, nr_cols_A);
GPU_fill_rand(d_B, nr_rows_B, nr_cols_B);
// Optionally we can copy the data back on CPU and print the arrays
cudaMemcpy(h_A,d_A,nr_rows_A * nr_cols_A * sizeof(float),cudaMemcpyDeviceToHost);
cudaMemcpy(h_B,d_B,nr_rows_B * nr_cols_B * sizeof(float),cudaMemcpyDeviceToHost);
std::cout << "A =" << std::endl;
print_matrix(h_A, nr_rows_A, nr_cols_A);
std::cout << "B =" << std::endl;
print_matrix(h_B, nr_rows_B, nr_cols_B);
// Multiply A and B on GPU
gpu_blas_mmul(d_A, d_B, d_C, nr_rows_A, nr_cols_A, nr_cols_B);
// Copy (and print) the result on host memory
cudaMemcpy(h_C,d_C,nr_rows_C * nr_cols_C * sizeof(float),cudaMemcpyDeviceToHost);
std::cout << "C =" << std::endl;
print_matrix(h_C, nr_rows_C, nr_cols_C);
//Free GPU memory
cudaFree(d_A);
cudaFree(d_B);
cudaFree(d_C);
// Free CPU memory
free(h_A);
free(h_B);
free(h_C);
return 0;
}
end

请先登录，再进行评论。

请先登录，再回答此问题。

回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Kernel Creation from MATLAB Code 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by