clock overhead blurb

clock_overhead.c

// clock_overhead
//  (code started from http://stackoverflow.com/questions/6814792/why-is-clock-gettime-so-erratic)
#include <time.h>
#include <stdio.h>
#include <stdint.h>
#include <stdlib.h>
#include <inttypes.h>

/* Compiled & executed with:
    gcc clock_overhead.c -O0 -lrt -o clock_overhead
    ./clock_overhead
*/

static inline uint64_t rdtsc()
{
#if defined(__i386__)
    uint64_t x;
    __asm__ volatile (".byte 0x0f, 0x31" : "=A" (x));
    return x;
#elif defined(__x86_64__)
    uint32_t hi, lo;
    __asm__ __volatile__ ("rdtsc" : "=a"(lo), "=d"(hi));
    return ((uint64_t)lo) | ((uint64_t)hi << 32);
#else
#error Unsupported architecture.
#endif
}

int main(int argc, char *argv[])
{
    int i;
    uint64_t tstart, tend;
    struct timespec tstart2, tend2;
    int N = (argc > 1) ? atoi(argv[1]) : 1000000;

#if defined(__i386__)
    const char arch[] = "32-bit";
#elif defined(__x86_64__)
    const char arch[] = "64-bit";
#else
    const char arch[] = "??-bit";
#endif

    printf("\n");
    printf("Running estimated overhead tests on %s.\n", arch);
    
    printf("\n");
    printf("CPU model from /proc/cpuinfo:\n");
    system("grep \"model name\" -m 1 /proc/cpuinfo");
   
    for (i = 0; i < 10; i++)
    {
        int n;

        printf("\n");

        clock_gettime(CLOCK_THREAD_CPUTIME_ID, &tstart2);
        tstart = rdtsc();
        for (n = 0; n < N; ++n)
        {
            uint64_t dummy;
 
            dummy = rdtsc();
            dummy = rdtsc();
            dummy = rdtsc();
            dummy = rdtsc();
            dummy = rdtsc();
            dummy = rdtsc();
            dummy = rdtsc();
            dummy = rdtsc();
            dummy = rdtsc();
            dummy = rdtsc();
        }
        tend = rdtsc();
        clock_gettime(CLOCK_THREAD_CPUTIME_ID, &tend2);

        printf("rdtsc         : %3" PRIu64 " ns",
               ((uint64_t) tend2.tv_sec * 1000000000 + (uint64_t) tend2.tv_nsec
                - ((uint64_t) tstart2.tv_sec * 1000000000
                   + (uint64_t) tstart2.tv_nsec)) / N / 10);
        printf(" %G cycles\n", (double)(tend - tstart)/N/10);

        clock_gettime(CLOCK_THREAD_CPUTIME_ID, &tstart2);
        tstart = rdtsc();
        for (n = 0; n < N; ++n)
        {
            struct timespec dummy2;

            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
            clock_gettime(CLOCK_MONOTONIC, &dummy2);
        }
        tend = rdtsc();
        clock_gettime(CLOCK_THREAD_CPUTIME_ID, &tend2);

        printf("clock_gettime : %3" PRIu64 " ns",
               ((uint64_t) tend2.tv_sec * 1000000000 + (uint64_t) tend2.tv_nsec
                - ((uint64_t) tstart2.tv_sec * 1000000000
                   + (uint64_t) tstart2.tv_nsec)) / N / 10);
        printf(" %G cycles\n", (double)(tend - tstart)/N/10);
    }
    return 0;
}