Re: architecture names (i386-linux, etc.)
On Wed, 1 Sep 1999, Alan Cox wrote:
> Date: Wed, 1 Sep 1999 23:24:41 +0100 (BST)
> From: Alan Cox <alan@lxorguk.ukuu.org.uk>
> To: quinlan@transmeta.com
> Cc: lsb-spec@lists.linuxbase.org
> Subject: Re: architecture names (i386-linux, etc.)
> Resent-Date: 1 Sep 1999 22:57:50 -0000
> Resent-From: lsb-spec@lists.linuxbase.org
> Resent-cc: recipient list not shown: ;
>
> > On the other issue, it might be helpful if there was some sort of
> > standard for x86 architecture names.
> >
> > Does "i686-linux" mean "only runs on a Pentium Pro or above" or does
> > "i686-linux" mean optimized for Pentium Pro, but still runs on i386?
> > I thought it meant the former, not the latter. Unfortunately, a some
> > number of packages just assign the number according to the system the
> > package was compiled on, not the target system.
of course this should be changed, because there are a lot of situation
that are matter of ambiguity. for example, if i upgrade system from
a pentium to a pentiumII, then i compile binutils, they goes
in /usr/i686-pc-linux-gnu, while the compilers i had
compiled before has files in /usr/i586-pc-linux-gnu, too many directories
for my tastes. i admit this is a very easy case to resolv, in many ways,
but for people who are news to any knid of Unix? disorder in directories
is the worst thing for system administration.
>
> Lets follow the intel recommendations. We then have
>
> i386 - all cpus
> i486 - requires i486 or higher (uses bswap)
>
> People following the rules and writing correct applications will not
> have a problem with just the two types. Intel AMD and Cyrix all tell people
> to use CPUID to check the mmx/3dnow!/kni/etc facilities are present and
> software fallback.
i agree, but i suggest ix86, for all cpus, so i386 could be used not as
a generic indication, but as a recommendation.
>
> We should mandate compliant packages do this. We should not mandate what a
> package is optimised for either IMHO, but what it runs on. "Recommended CPU"
> is a marketing issue not a standards one.
>
> Alan
>
This is an important point. actually there are too many different x86 cpus
to assume "recommended CPU" as a standards issue. a part of that, it is a
matter of phylosophical view. a package should ever declare what it runs
on, (from this cpu upper), but would be fine for a lot of people to
be able to know also what it is optimizzed for. in my opinion, just my
two cents, everyone will desagree with everyone about that point.
a good system administrator would prefer compile sources to gain an eavy
optimizzation
for the machine he manages, and so it is normal that the name declare
this situation, but most of the distribution are optimizzed almost
for i486, not more, so that the packages would run also on 386, so the
name for pre-compiled distributions packets should declare that.
it's a matter of mental order.
Luigi Genoni
Reply to: