> Am 25.07.2025 um 17:16 schrieb Robin Dapp <rdapp....@gmail.com>:
>
>
>>
>> I'm not sure whether handling this case as part of VMAT_STRIDED_SLP is
>> wise. IIRC we do already choose VMAT_GATHER_SCATTER for some
>> strided loads, so why not do strided load/store handling as part of
>> gather/scatter handling?
>
> Now that we can deal with gather/scatter misalignment I think we can come
> back to this.
>
> IMHO strided loads are an ok fit for STRIDED_SLP because of the support for
> composing larger elements from individual bytes. To my understanding this is
> the only part of the vectorizer where we attempt these composition types.
>
> Regular gather/scatter handling doesn't have this capability yet and we
> already use strided loads/stores if we can determine the gather/scatter index
> is simple.
>
> Would maybe refactoring this strategy help with making it suitable for
> strided loads/stores?
That would definitely be nice to have for both gather and stride loads
Richard
>
> --
> Regards
> Robin
>
