[PATCH v11 10/29] string: Improve generic stpcpy

[Richard Henderson]

On 2/1/23 09:47, Noah Goldstein wrote:
>> +static __always_inline char *
>> +stpcpy_unaligned_loop (op_t *restrict dst, const op_t *restrict src,
>> +                      uintptr_t ofs)
>> +{
> 
> Is the perf cost of unaligned stores greater than merging?

Generically, yes.  Some hosts have no hw unaligned stores at all without even fallback to 
trap and emulate, and compiler-generated unaligned stores devolve to byte-by-byte.



r~