topic Hi Sem, in Intel® oneAPI Math Kernel Library

Problem with mkl fft dynamic array

Semen_K_ — Mon, 03 Apr 2017 12:22:06 GMT

Hi, I want to use the mklfft with multithreading, but I faced the problem. I got an error - Segmentation fault (core dumped) even when I just created the simple 2D dimension dynamic array. Also, I have checked the multithread example with static array, it worked perfectly, but when I created dynamic array, it gave me an error. Could you tell me what is wrong I am doing, please?

This is my code:

#include "mkl_dfti.h"

#include <omp.h>

#include <math.h>

#include <complex.h>

#include <float.h>

#include <sys/time.h>

int main(int argc, char *argv[])

{

int k, j, N = atoi(argv[1]);

MKL_Complex8 **data = malloc(N*sizeof(MKL_Complex8*));

for(k=0; k<N; k++){

data = malloc(N*sizeof(MKL_Complex8));

}

printf(

"Before FFT: N %d.\n",

N);

MKL_LONG len[2] = {N, N};

DFTI_DESCRIPTOR_HANDLE FFT;

int th;

printf("Create...\n");

DftiCreateDescriptor (&FFT, DFTI_SINGLE, DFTI_COMPLEX, 2, len);

DftiCommitDescriptor (FFT);

printf("Execute...\n");

struct timeval start, end;

gettimeofday(&start, NULL);

DftiComputeForward (FFT, data);

DftiFreeDescriptor (&FFT);

gettimeofday(&end, NULL);

double tstart = start.tv_sec + start.tv_usec/1000000.;

double tend = end.tv_sec + end.tv_usec/1000000.;

double t_sec = (tend - tstart);

int S = N*N;

double speed = (double) 5*S*log2(S)/t_sec*1e-6;

printf("Time: %0.6f, Speed: %0.6f\n", t_sec, speed);

for(j=0; j<N; j++) {

free(data);

}

free(data);

return 0;

}

Dear customer,

Zhen_Z_Intel — Wed, 05 Apr 2017 08:45:35 GMT

Dear customer,

The MKL FFT function for execution's required input datatype is pointer for complex/floating array, not address of pointer. You probably need define the data as MKL_Complex8 * which size is N*N. Please do following work:

MKL_Complex8 *data = (MKL_Complex8*)malloc(N*N*sizeof(MKL_Complex8));
/*MKL_Complex8 **data =(MKL_Complex8 **) malloc(N*sizeof(MKL_Complex8*));
    for(k=0; k<N*N; k++){
        data = (MKL_Complex8*)malloc(N*sizeof(MKL_Complex8));
    }*/
....

/*for(j=0; j<N; j++) {
    free(data);
}*/
free(data);

You could refer FFT sample to write.

Best regards,
Fiona

Dear Fiona,

Semen_K_ — Wed, 05 Apr 2017 09:54:56 GMT

Dear Fiona,

Thanks for your reply. I have tried it, and it works, but I need to do my computations in parallel, so, I need to each thread has it's own data. I got it that I have to pass array (not multidim array) to the function, but I need to find the way to parallelize it. For instance,

for (int i=0; i<numthreads; i++)
       DftiComputeForward (FFT, x);

Now, with one pointer, if I do like &x[i*N/numthreads*N] it is not working.

Thanks,
Sem

Hi Sem,

Zhen_Z_Intel — Thu, 06 Apr 2017 04:01:48 GMT

Hi Sem,

Please refer this link of sample for parallel computing with one common descriptor/multiple descriptor. Thanks.

Best regards,
Fiona