output

0.900109s
1.508606s

ping-pong1.c

#include <time.h>
#include <stdio.h>
#include <stdlib.h>
#include <pthread.h>
#include <string.h>
#include <assert.h>

#define tic() do { struct timespec ts_start, ts_end; clock_gettime(CLOCK_MONOTONIC, &ts_start)
#define toc() clock_gettime(CLOCK_MONOTONIC, &ts_end); \
              printf("%lfs\n", (ts_end.tv_sec - ts_start.tv_sec) + (double)(ts_end.tv_nsec - ts_start.tv_nsec)/1e9); } \
              while (0)

#define NUM_THREADS 4
#define N 536870912 

void* foo(void *);
void* bar(void *);

int* num;
int s[NUM_THREADS];

int main()
{
    pthread_t threads[NUM_THREADS];
    int ans, ANS = 0;
    num = (int*) malloc(sizeof(int) * N);

    srand(42);

    for (int i = 0; i < N; ++i) {
        num[i] = rand() % 100;
        ANS += num[i];
    }

    memset(s, 0, sizeof(s));
    ans = 0;
    tic();
    for (int i = 0; i < NUM_THREADS; ++i)
        pthread_create(threads + i, NULL, foo, s + i);
    for (int i = 0; i < NUM_THREADS; ++i) {
        pthread_join(threads[i], NULL);
        ans += s[i];
    }
    toc();
    assert(ans == ANS);

    memset(s, 0, sizeof(s));
    ans = 0;
    tic();
    for (int i = 0; i < NUM_THREADS; ++i)
        pthread_create(threads + i, NULL, bar, s + i);
    for (int i = 0; i < NUM_THREADS; ++i) {
        pthread_join(threads[i], NULL);
        ans += s[i];
    }
    toc();
    assert(ans == ANS);

    return 0;
}

void* foo(void* ptr)
{
    int* r = (int*) ptr;

    for (int i = (r - s); i < N; i += NUM_THREADS)
        *r += num[i];
    return NULL;
}

void* bar(void* ptr)
{
    int* r = (int*) ptr;
    int idx = r - s;
    int block = N/NUM_THREADS;
    int start = idx * block, end = start + block;

    for (int j = 0; j < NUM_THREADS; ++j)
        for (int i = start+j; i < end; i+=NUM_THREADS)
            *r += num[i];
    return NULL;
}