Re: Cost of handling misaligned access

On Fri, 14 Feb 2025 21:14:11 +0000, BGB wrote:

On 2/13/2025 1:09 PM, Marcus wrote:

-------------

>
The problem arises when the programmer *deliberately* does unaligned
loads and stores in order to improve performance. Or rather, if the
programmer knows that the hardware supports unaligned loads and stores,
he/she can use that to write faster code in some special cases.
>

>
Pretty much.
>
>
This is partly why I am in favor of potentially adding explicit keywords
for some of these cases, or to reiterate:
   __aligned:
     Inform compiler that a pointer is aligned.
     May use a faster version if appropriate.
       If a faster aligned-only variant exists of an instruction.
       On an otherwise unaligned-safe target.
   __unaligned: Inform compiler that an access is unaligned.
     May use a runtime call or similar if necessary,
       on an aligned-only target.
     May do nothing on an unaligned-safe target.
   None: Do whatever is the default.
     Presumably, assume aligned by default,
       unless target is known unaligned-safe.

It would take LESS total man-power world-wide and over-time to
simply make HW perform misaligned accesses.