#include <mach/mach_time.h>
#include <time.h>
#include <stdio.h>

#define DUFF_DEVICE_8(aCount, aAction)	\
do {	\
int count_ = (aCount);	\
int times_ = (count_ + 7) >> 3;	\
switch (count_ & 7){	\
case 0: do { aAction;	\
case 7: aAction;	\
case 6: aAction;	\
case 5: aAction;	\
case 4: aAction;	\
case 3: aAction;	\
case 2: aAction;	\
case 1: aAction;	\
} while (--times_ > 0);	\
}	\
} while (0)

void mach_absolute_difference(uint64_t end, uint64_t start, struct timespec *tp) {
        uint64_t difference = end - start;
        static mach_timebase_info_data_t info = {0,0};

        if (info.denom == 0)
                mach_timebase_info(&info);

        uint64_t elapsednano = difference * (info.numer / info.denom);

        tp->tv_sec = elapsednano * 1e-9;
        tp->tv_nsec = elapsednano - (tp->tv_sec * 1e9);
}

void *simplememset(void *_p, unsigned v, unsigned count)
{
    unsigned char *p = _p;
    while(count-- > 0) *p++ = v;
    return _p;
}

void *duffmemset(void *_p, unsigned v, unsigned count)
{
  register n = (count + 7) / 8;      /* count > 0 assumed */
  unsigned char *p = _p;

  switch (count % 8)
  {
    case 0:        do {*p++ = v;
    case 7:              *p++ = v;
    case 6:              *p++ = v;
    case 5:              *p++ = v;
    case 4:              *p++ = v;
    case 3:              *p++ = v;
    case 2:              *p++ = v;
    case 1:              *p++ = v;
                   } while (--n > 0);
    }

    return _p;
}

int main(int argc, char *argv[])
{
  uint64_t start,end;
  struct timespec tp;
  int rep;
  int i;
  int arraylen = atoi(argv[1]);
  unsigned *array, *tmp;
  printf("Array Length: %d\n", arraylen);

  for( rep = 0; rep < 5; rep++ )
  {
    array = ( unsigned * )malloc( arraylen * sizeof( unsigned ) );
    start = mach_absolute_time();
    memset(array, 0, arraylen * sizeof( unsigned ));
    end = mach_absolute_time();
    mach_absolute_difference(end, start, &tp);
    if( rep > 0 )
      printf("Native: %lu seconds, %lu nanoseconds\n", tp.tv_sec, tp.tv_nsec);
    free( array );

    array = ( unsigned * )malloc( arraylen * sizeof( unsigned ) );
    start = mach_absolute_time();
    duffmemset(array, 0, arraylen);
    //tmp = array;
    //DUFF_DEVICE_8( arraylen, *tmp++ = 0 );
    end = mach_absolute_time();
    mach_absolute_difference(end, start, &tp);
    if( rep > 0 )
      printf("Duff: %lu seconds, %lu nanoseconds\n", tp.tv_sec, tp.tv_nsec);
    free( array );

    array = ( unsigned * )malloc( arraylen * sizeof( unsigned ) );
    start = mach_absolute_time();
    simplememset(array, 4, arraylen * sizeof( unsigned ) );
    end = mach_absolute_time();
    mach_absolute_difference(end, start, &tp);
    if( rep > 0 )
      printf("Simple: %lu seconds, %lu nanoseconds\n", tp.tv_sec, tp.tv_nsec);
    free( array );

    printf("\n");
  }

  return 0;
}