[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: subarchitectures (was Re: What to do with optimization flags ? )


Ben Collins <bcollins@debian.org> writes:

> On Thu, Dec 07, 2000 at 12:14:53PM -0500, Camm Maguire wrote:
> > Greetings, and thank you so much for this helpful information!  
> > 
> > 1) I notice you reference 'apic'.  Do you happen to know to what that
> >    refers?  The ones I recognize are mmx,xmm,and amd3d.
> > 
> > 2) There are some small differences between k6 amd3d and athlon amd3d.
> >    Do you know which is referred to here?
> The flags correspond to output from /proc/cpuinfo (the "flags" field).
> Like on my Celeron laptop I have:
> %cat /proc/cpuinfo | grep flags
> flags           : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 pn mmx fxsr

I'm getting around to this now, and had a question:  What is the flag
that indicates SSE2 (I take it xmm -> SSE1).  I've looked through the
kernel source, and cannot find it.


> > 3) We've already had a brief discussion about this, but to fill you
> >    in, the major two alternatives are to either have several package
> >    binaries for each sub-arch, or to have all libs in one binary
> >    package.  In either case, it sounds like a good ide to use the
> >    right subdir.  I was wondering if you had a suggestion between
> >    these two alternatives.
> For libc6 I use seperate packages, but that is mainly because I don't want
> to force everyone to install all of these libs (since everyone has to have
> libc6 whether they want it or not :)
> The downside to seperate packages, is the increase in the Packages file,
> and the number of available packages in the dist.
> > 4) Most importantly, it appears that atlas cannot cross-compile,
> >    i.e. the compiled code must *run* on the compilation machine.  From
> >    what I can see, this makes it impossible to autobuild fully
> >    optimized version(s) of this package given the current machines at
> >    Debian's disposal.  I can set the package up to autobuild a generic
> >    x86 lib, like it does currently, which will autobuild successfully.
> >    But I could also produce a fully optimized binary package covering
> >    p3,p2,k6,k7 given the machines available here.  My question: is
> >    there anyway to get such a binary package to override the
> >    auto-built binary packages in the distribution?
> That's a downfall in atlas, IMO. There should be a way to compile for a
> CPU, without actually having that CPU. You will most likely have to hack
> the atlas build process to get this done.
> -- 
>  -----------=======-=-======-=========-----------=====------------=-=------
> /  Ben Collins  --  ...on that fantastic voyage...  --  Debian GNU/Linux   \
> `  bcollins@debian.org  --  bcollins@openldap.org  --  bcollins@linux.com  '
>  `---=========------=======-------------=-=-----=-===-======-------=--=---'

Camm Maguire			     			camm@enhanced.com
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah

Reply to: