Using PPL with CArray and Quad tree for parallel processing

InShort 2 issues:
1. use ppl with MFC so that each branch of a quad tree can be processed independently, store the results in a shared static CArray
2. use ppl with MFC so that each node of the quad tree when processed can also write to a shared static array of data

I have a quad tree, and what I want to do is have either PPL, or OMP to do each quad to execute down it's own branch doing there respective quad

so I'm using MFC and I have an CArray that stores each final leaves of each quad tree branch.

I was trying to do it in PPL or OMP but not very successful. It works fine non parallel it just takes a while on large data sets.

I'm new to this idea, but hopefully someone that is savy with PPL can tell me how to do MFC.

so:

static CArray <PtrArray,CNodes> m_NodesPtrArray;

QuadTree::BuildQuadTree(Triangle atriangle[], double width, double length)
{

depth=0;
root->Rectangle.SetSize(width,length);
CArray<int,int>Node_indexes_of_Triangles;//this has total cnt as well as 0->N triangle indexes
root -> SplitNode(atriangle, Node_indexes_of_Triangles,depth);
}

CNode::SplitNode(Triangle * Array, CArray<int,int)*pArrayIndexs, int depth)
{
depth++;

if(depth=0)
{// run in parallel
  CPoints apt;
  GetCenterofQuad( &apt)
  CArray <int,int>QArray1,QArray2,QArray3, QArray4;
  GetTrianglesPerNewQuads(apt,&QArray1,&QArray2,&QArray3, &QArray4);

  parallel_invoke(
          [&]{SplitNode(atriangle,QArray1, depth);},
          [&]{SplitNode(atriangle,QArray2, depth);},
          [&]{SplitNode(atriangle,QArray3, depth);},
          [&]{SplitNode(atriangle,QArray4, depth);}
  );
}
else if(depth<4)
{
  CPoints apt;
  GetCenterofQuad( &apt)
  CArray <int,int>QArray1,QArray2,QArray3, QArray4;
  GetTrianglesPerNewQuads(apt,&QArray1,&QArray2,&QArray3, &QArray4);

  SplitNode(atriangle,QArray1, depth);
        SplitNode(atriangle,QArray2, depth);
        SplitNode(atriangle,QArray3, depth);
        SplitNode(atriangle,QArray4, depth);
}
else
{//at final depth
  //write this node to PtrArray for processing later...
  this->SetIndex(pArrayIndexs);
  m_NodesPtrArray.Add(this);
  return;
}
}

I have another obvious issue, is how do you then process each node after this in parallel AND also have an array that each thread can write to???