gccのprefetch

FrontPage

• moves the data from memory to cache in order to speed up the memory access
• https://gcc.gnu.org/onlinedocs/gcc-5.3.0/gcc/Other-Builtins.html
• Example from here

  for (i = 0; i < n; i++)
  {
      a[i] = a[i] + b[i];
      __builtin_prefetch (&a[i+j], 1, 1);
      __builtin_prefetch (&b[i+j], 0, 1);
      /* ... */
  }

• __builtin_prefetch takes 3 parameters, which later 2 are optional
• the first parameter is the data to be prefetched
• the second parameter is 0 or 1, compile time constant
  • 0 is to read the data (which is default)
  • 1 is to write
• the third parameter is locality 0,1,2 or 3 and has to be compile time constant
  • 0 is no temporal locality. it will be purged after access
  • 3 is strong temporal locality. It'll try to keep the data in cache as much as possible(default)