OMP task final

Hello, I have just started to learn omp tasks and have a question about final clause. I compiled the following code based on the example from https://software.intel.com/sites/products/documentation/doclib/iss/2013/...  with Intel C++ 15.0.0.108 and ran it on simple binary tree of level 3.

#include <iostream>
#include <omp.h>

const int MAX_LEVEL = 3;
using namespace std;

struct node {
	struct node *left;
	struct node *right;
	int nodeNumber;
};

void process(node * pNode)
{
#pragma omp critical
	{
	cout<<omp_get_thread_num()<<""<<pNode->nodeNumber<<endl;
	}
}

void traverse( struct node *p, int lev, int finLevel ) {
	if (p->left)
#pragma omp task  final(1) // p is firstprivate by default
		traverse(p->left,lev+1,finLevel);
	if (p->right)
#pragma omp task final(1)// p is firstprivate by default
		traverse(p->right,lev+1, finLevel);
	process(p);
}

void allocateNodes(node **p, int level, int *nodeNumber) {
	*p = new node;
	(*p)->nodeNumber = *nodeNumber;
	if (level < MAX_LEVEL)
	{
		(*nodeNumber)++;
		allocateNodes(&(*p)->left, level+1, nodeNumber);
		(*nodeNumber)++;
		allocateNodes(&(*p)->right, level+1, nodeNumber);
	}
	else
	{
		(*p)->left = NULL;
		(*p)->right = NULL;
	}
}

int main()
{
	node *p;
	int nodeNumber = 0;
	allocateNodes(&p,0,&nodeNumber);
#pragma omp parallel
#pragma omp single
	traverse(p,0,-1);
}

And get strange output:

0  0
1  1
2  8
0  9
1  5
2  12
0  11
1  7
2  14
0  10
1  6
2  13
0  2
1  3
2  4

But If I compile the same code with g++ I get another output:

2  3
0  0
2  4
0  10
2  2
0  11
2  6
0  9
2  7
2  5
0  13
2  1
0  14
0  12
0  8