Re: [boost] Boost SIMD beta release

Subject: Re: [boost] Boost SIMD beta release
From: Shakti Misra (shakti.misra.study_at_[hidden])
Date: 2012-12-20 11:22:49

Nice work. Just a quick question: In case my arch do not support a
instruction, do you have a default implementation? Like a fallback?

On Thu, Dec 20, 2012 at 7:39 PM, Andrey Semashev
<andrey.semashev_at_[hidden]>wrote:

> On Thu, Dec 20, 2012 at 5:49 PM, Thorsten Ottosen
> <thorsten.ottosen_at_[hidden]> wrote:
> > On 20-12-2012 12:52, Mathias Gaunard wrote:
> >
> >> You can choose which extension to generate code for at compile time.
> >>
> >> The approach we recommend is to compile various versions of your
> >> function with different settings, and then choose the right one at
> >> runtime depending on the host capabilities.
> >> We provide functions to easily check whether an extension is supported.
> >
> >
> > That seems cool. Can illegal instructions be in a binary as long as they
> are
> > not exeuted?
>
> Yes, no problem with that.
>
> The suggested approach has a nasty potential problem though. You have
> to be extra careful so that no common inline functions are compiled in
> different translation units with different compiler settings.
> Otherwise you may have ODR violation and it is unspecified which
> version of such functions end up in the compiled binary.
>
> _______________________________________________
> Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost
>