c++ - How to allocate Array of Pointers and preserve them for multiple kernel calls in cuda -

i trying implement algorithm in cuda , need allocate array of pointers point array of structs. struct is, lets say:

    typedef struct {        float x, y;      } point; 

i know if want preserve arrays multiple kernel calls have control them host, right? initialization of pointers must done within kernel. more specific, array of struct p contain random order of cartesian points while dev_s_x sorted version x coordinate of points in p.

i have tried with:

__global__ void test( point *dev_p, point **dev_s_x) {     unsigned int tid = threadidx.x + blockidx.x * blockdim.x;      dev_p[tid].x = 3.141516;     dev_p[tid].y = 3.141516;     dev_s_x[tid] = &dev_p[tid];    ... } 

and:

 int main( void ) {      point *p, *dev_p, **s_x, *dev_s_x;      p   = (point*)  malloc (n * sizeof (point) );      s_x = (point**) malloc (n * sizeof (point*));       // allocate memory on gpu      cudamalloc( (void**)  &dev_p,   n * sizeof(point) );      cudamalloc( (void***)  &dev_s_x, n * sizeof(point*));       // copy array p gpu      cudamemcpy( dev_p, p,  n * sizeof(point),  cudamemcpyhosttodevice);      cudamemcpy( dev_s_x,s_x,n * sizeof(point*), cudamemcpyhosttodevice);       test <<<1, 1 >>>( dev_p, &dev_s_x);         ...      return 0; } 

which leads many

first-chance exception @ 0x000007fefcc89e5d (kernelbase.dll) in test_project_cuda.exe: microsoft c++ exception: cudaerror_enum @ memory location 0x0020f920.. critical error detected c0000374

am doing wrong in cudamalloc of array of pointers or else? usage of (void***) correct? use example dev_s_x[tid]->x or dev_s_x[tid]->y within kernels pointing device memory addresses. feasible? in advance

dev_s_x should declared point ** , should passed kernel value (i.e. test <<<1, 1 >>>(dev_p, dev_s_x);).

putting 1 side, describe sounds natural fit thrust, give simpler memory management strategy , access fast sort routines.