#include <iostream>
#include <cstdlib>
#include <cmath>
#include <xmmintrin.h>
using namespace std;

typedef float real;

const int TEN_MILLION = 10000000;

void process(real *x, real *y, real *z, int size) {
	#pragma omp parallel for simd schedule(static,10) 
	for (int i=0; i<size; ++i) {
		z[i] = (sin(x[i]) - cos(y[i])) / (sqrt(x[i] * x[i] + y[i] * y[i]));
	}
}

int main() {
	real *a, *b, *c;
	
	a = (real *) _mm_malloc( TEN_MILLION * sizeof(real), 16);
	b = (real *) _mm_malloc( TEN_MILLION * sizeof(real), 16);
	c = (real *) _mm_malloc( TEN_MILLION * sizeof(real), 16);
	
	for (int i=0; i<TEN_MILLION; ++i) {
		a[i] = i % 1000;
	}
	for (int i=0; i<TEN_MILLION; ++i) {
		b[i] = i % 133;
	}
	
	int iterations = 0;
	for (iterations = 0; iterations < 10'000; ++iterations) {
	
		process(a,b,c,TEN_MILLION);
		
	}
	cout << "iterations=" << iterations << endl;
	
	_mm_free(a);
	_mm_free(b);
	_mm_free(c);
	
	return EXIT_SUCCESS;
}