Help with simple array question

[wmarquez]

Hello! I have a question that I think might be simple to answer but I haven't found any documentation anywhere to confirm my suspicions. I pasted only the relevant portions of my code below to the question that I'm asking. My program is supposed to multiply two square matrices.

The question that I have is about an "error: incompatible types in assignment" that I get when I try to assign suba, subb, subc with the corresponding portions of a, b, and c. Is this because I'm using variables v and w? Also, just to make sure that I have the right concept, if I assign the top left corner of a matrix to a "submatrix" then I'm just assigning a pointer (such as subb) to that start at the specified position of the big matrix, right?

Thanks in advance for the help! It is IMMENSELY appreiciated

struct threads
{
  pthread_t id; //The thread id to use in functions
  int n;    //size of block
  float **suba; //Sub-matrix a
  float **subb; //Sub-matrix b
  float **subc; //Sub-matrix c
};

int main( int argc, char* argv[ ] )
{
  int n; // dimension of the matrix 
  int p; // number of threads in the program
  float *sa, *sb, *sc; // storage for matrix A, B, and C
  float **a, **b, **c; // 2-d array to access matrix A, B, and C
  int i, j;
  struct threads* t;

  t = ( struct threads* ) malloc( p * sizeof( struct threads ) );

  int x = -1;
  int z;
  for( z = 0; z < p; z++ )
  {
    t[ z ].n = n / sqrt( p );
    if( fmod( z, sqrt( p ) ) == 0 )
      x++;
    int w = ( int )( x * n / sqrt( p ) );
    int v = ( int )( fmod( z, sqrt( p ) ) * n / sqrt( p ) );
    t[z].suba = a[ w ][ v ];
    t[z].subb = b[ w ][ v ];
    t[z].subc = c[ w ][ v ];

    //pthread_create( &t[ z ].id, 0, threadWork, &t[ z ] );
  }

  return 0;
}

[deceptikon]

Let's use t[z].suba = a[ w ][ v ]; as the example. Both suba and a have the same type (float**), so why are you subscripting a when trying to assign directly to suba? I suspect you're trying to copy the whole matrix:

t[z].suba = a;

This of course has other problems because all of the suba's in t will refer to the same matrix due to pointer aliasing, but fixing that involves allocating memory and changing how you copy the matrix from a direct pointer assignment to an element by element copy.