Page MenuHomec4science

pi_race_condition.cc
No OneTemporary

File Metadata

Created
Mon, Nov 11, 03:24

pi_race_condition.cc

/*
This exercise is taken from the class Parallel Programming Workshop (MPI,
OpenMP and Advanced Topics) at HLRS given by Rolf Rabenseifner
*/
#include <chrono>
#include <cstdio>
#include <cmath>
#include <omp.h>
using clk = std::chrono::high_resolution_clock;
using second = std::chrono::duration<double>;
using time_point = std::chrono::time_point<clk>;
inline int digit(double x, int n) {
return std::trunc(x * std::pow(10., n)) - std::trunc(x * std::pow(10., n - 1)) *10.;
}
inline double f(double a) { return (4. / (1. + a * a)); }
const int n = 10000000;
int main(int /* argc */ , char ** /* argv */) {
int i;
double dx, x, sum, pi;
auto nthreads = omp_get_max_threads();
auto t1 = clk::now();
/* calculate pi = integral [0..1] 4 / (1 + x**2) dx */
dx = 1. / n;
sum = 0.0;
#pragma omp parallel for shared(sum) private(x)
for (i = 0; i < n; i++) {
x = 1. * i * dx;
sum = sum + f(x);
}
pi = dx * sum;
second elapsed = clk::now() - t1;
std::printf("computed pi = %.16g\n", pi);
std::printf("wall clock time (chrono) = %.4gs\n", elapsed.count());
std::printf("using n threads = %.d\n", nthreads);
for(int d = 1; d <= 15; ++d) {
std::printf("%d", digit(pi, d));
}
return 0;
}

Event Timeline