Exophase (exophase@gmail.com) on 12/21/11 wrote:
---------------------------
>Michael S (already5chosen@yahoo.com) on 12/21/11 wrote:
>---------------------------
>>As to "Scaled, 12-bit, unsigned immediate offset", it seem to promote rather unusual
>>strategy for sorting fields within big structures: bytes first, then half-words,
>>then words. That's the opposite to what I accustomed to.
>>
>
>I hadn't really thought about that, that's a good point. I'm used to sorting by
>largest alignment first too, probably for the same reason you are. If posed with
>a decision regarding this I'll probably only change the sorting if it's needed to keep access to the smaller stuff.

In terms of padding overhead there is no difference between largest or smallest first. In any case for structures this won't matter on ARM64 as the smallest offset is a whopping 4KB! On Thumb-1 however it matters as the maximum offset for 8/16/32-bit loads is just 31/62/124. So smallest first has been best for a while.

Wilco