[csw-maintainers] Why do I get V8+ binaries?

[James Lee]

On 06/09/09, 21:19:45, Dagobert Michelsen <dam at opencsw.org> wrote regarding 
Re: [csw-maintainers] Why do I get V8+ binaries?:

> > sparc, the binary which is created is a V8+ binary. So my questions
> > are:
> >
> > (1) Why it becomes a V8+? I cannot find a reason for that.
> > (2) Is this a critical thing or can I ignore it?

> It is sort of critical as the binaries are then no longer runnable on
> all
> machines supported by Solaris 8. However, as we officially deprecated
> Solaris 8 this is not really bad, but still a Good Thing(tm) to do.

Solaris 9 runs on sun4m so the Solaris 8 point is invalid.  We will
continue to support 9 for some time so the sun4m question remains.

Does sun4m matter?  It seems to matter to a few people and although
they are only small number it is 100% critical to them.

I do my bit to run a living computer museum but I am long past running
a sun4m machine for anything other than a toy or test.  I can just see
why a stand alone server might be better on a Sparc20 as my SS20
(125MHz, 512GB RAM) uses 58 Watts idle whereas my Blade2000 (1.2GHz,
8GB RAM) use 185W idle.  Of course the Blade 2000 can do what a the
SS20 can in one of its corners with no one noticing plus 10 times more
but this assume I have the faster machine running and have more to do.
Why anyone chooses to run GUI apps on such slow hardware is beyond me.
However because we run an integrated system long dependency chains
dictate the base arch, i.e. if one package fails the system fails.

Finally it's worth pointing out that in most cases sun4m (code not
chip) gives the same performance as sun4u (sparcv8plus+vis) and where
it differs and matters we have the isaexec and $ISALIST mechanisms.
Consider also that for many programs speed is not important and small
size might be the goal (-xO3 -xspace), e.g. background daemons and
network limited.



James.