Re: HAVE_FAST_UNALIGNED_ACCESS

[Arsen Arsenović]

Mattias Engdegård <mattiase@acm.org> writes:

> 30 mars 2023 kl. 14.18 skrev Arsen Arsenović <arsen@aarsen.me>:
>
>> As an example, the compiler is free to, for instance, vectorize a loop,
>> emitting instructions that very much have alignment checking even on
>> x86 (the loop in question is very much parallelizable and vectorizable,
>> as it feels like a textbook example of such operations).
>
> If only. Then we wouldn't have this problem, would we.

Free to, not required to ;)

Thanks for pushing your patch.  I'm still unsure if the machine specific
knowledge here helps, but this is an improvement nonetheless.

Have a lovely day.
-- 
Arsen Arsenović

signature.asc
Description: PGP signature