Commented magma_dports_gpu

./dtrsm 

1.000000 0.000000 0.000000 
-0.400000 1.000000 0.000000 
-0.080000 -0.300000 1.000000 

1.000000 0.000000 0.000000 
-0.400000 1.000000 0.000000 
-0.080000 -0.300000 1.000000 

dtrsm.c

#include <stdio.h>
#include <cuda.h>
#include <cuda_runtime.h>
#include <cublas_api.h>
#include <magma.h>
#include <clapack.h>

void dpotrf(double *A, double *B, double *C, int N, int gpu) {

    int info = 0;
    if(gpu) {

        double *d_A, *d_B;
        cudaMalloc((void**)&d_A, N * N * sizeof(double));
        cudaMalloc((void**)&d_B, N * N * sizeof(double));
        
        cublasSetMatrix(N, N, sizeof(*A), A, N, d_A, N);
        cublasSetMatrix(N, N, sizeof(*B), B, N, d_B, N);

        magma_dpotrs_gpu('U', N, N, d_A, N, d_B, N, &info);
        
        cublasGetMatrix(N, N, sizeof(*d_B), d_B, N, C, N);

        cudaFree(d_A);
        cudaFree(d_B);

    } else {
       info = clapack_dpotrs(CblasColMajor, CblasUpper, N, N, A, N, C, N);    
    }

    if(info < 0) 
        printf("Error: Illegal argument %d in 'dpotrf'", -1 * info);
}

void print_mat(double *A, int M, int N) {
    
    for(int i=0; i<M*N; i++) {
        if (i % N == 0) 
            printf("\n");
        printf("%f ",A[i]);
    }
    printf("\n");
}


int main(void)
{
    int N = 3;
        
    double A[9] = {1.0, 0.0, 0.0, 
                   0.4, 1.0, 0.0, 
                   0.2, 0.3, 1.0 };
    double B[9] = {1,0,0,
                   0,1,0,
                   0,0,1};
    double C[9];
    
    double *d_A, *d_B, *d_C;
    
    cudaMalloc((void**)&d_A, N * N * sizeof(double));
    cudaMalloc((void**)&d_B, N * N * sizeof(double));
    
    cublasSetMatrix(N, N, sizeof(double), A, N, d_A, N);
    cublasSetMatrix(N, N, sizeof(double), B, N, d_B, N);
    
    double alpha=1.0;
    magmablas_dtrsm('L','U','N','N', N, N, alpha, d_A, N, d_B, N);
    
    cublasGetVector(N * N, sizeof(double), d_B, 1, C, 1);
    
    cudaFree(d_A);
    cudaFree(d_B);
    
    print_mat(C,N,N);
    
    
    cblas_dtrsm(CblasColMajor, CblasLeft, CblasUpper, CblasNoTrans, CblasNonUnit, N, N, alpha, A, N, B, N);    
    //clapack_dtrsm("L", "U", "N", "N", &N, &N, &alpha, A, &N, B, &N);
    
    print_mat(B,N,N);
    
    return 0;
}

gistfile1.sh

gcc -I/usr/local/cuda/include -I/usr/local/magma/include -I/usr/include/atlas -std=gnu99 -O3 -g -c dtrsm.c -o dtrsm.o
gcc -o dtrsm dtrsm.o -L/usr/local/cuda/lib64 -L/usr/local/magma/lib -lcuda -lcudart -lcublas -lmagmablas -lmagma -llapack -lblas

Uncommented magma_dpotrs_gpu

./dtrsm 

0.000000 0.000000 0.000000 
0.000000 0.000000 0.000000 
0.000000 0.000000 0.000000 

1.000000 0.000000 0.000000 
-0.400000 1.000000 0.000000 
-0.080000 -0.300000 1.000000