Easy Duff Unrolling

duff.h

// Set ITER to be the code to execute at each iteration.
// Only use this when you can't guarantee your number of elements is a multiple of 8 (or some higher power of 2)
// If you want to save some time, you can initialize your accumulator with the first entry of the set and decrement len.

#define DO_DUFF(len, ITER) \
    do { \
        if(len) {\
            std::uint64_t loop = (len + 7) >> 3;\
            switch(len & 7) {\
                case 0: do {\
                    ITER; [[fallthrough]];\
                    case 7: ITER; [[fallthrough]]; case 6: ITER; [[fallthrough]]; case 5: ITER; [[fallthrough]];\
                    case 4: ITER; [[fallthrough]]; case 3: ITER; [[fallthrough]]; case 2: ITER; [[fallthrough]]; case 1: ITER;\
                } while (--loop);\
            }\
        }\
    } while(0)