Am Mittwoch, den 10.07.2013, 10:26 +1200 schrieb Andre Renaud:
> typedef struct multiple_size_t {
>     size_t d[4];
> } multiple_size_t;

why not have it

typedef size_t _multiple_size[4];

the wrapping into struct just doesn't serve much purpose, I think.

Then for your implementation, the commonly used trick would be to have
two "slow" phases for misalignment. One at the start for the first
bytes up to the next valid alignment boundary, do the "fast" copy for
the aligned part, and then handle the last bytes in another slow
phase. For small things to copy this adds a bit of arithmetic and a
conditional.

Jens

-- 
:: INRIA Nancy Grand Est :: http://www.loria.fr/~gustedt/   ::
:: AlGorille ::::::::::::::: office Nancy : +33 383593090   ::
:: ICube :::::::::::::: office Strasbourg : +33 368854536   ::
:: ::::::::::::::::::::::::::: gsm France : +33 651400183   ::
:: :::::::::::::::::::: gsm international : +49 15737185122 ::