Great, i686 without SSE(2) seems OK to me. I even wonder why we did not
> Chris Adams (cmadams hiwaay net
> > Removing support for still-functional hardware is a trademark of
> > Microsoft, not Linux.
> > I'd also argue that doing another full rebuild of the OS for a 1%
> > performance gain on a single architecture is not a particularly
> > production use of resources.
> The 1% comes from i586 -> i686; SSE2 would be additional on top of
> that. But given the vehement opposition, I can see dropping the SSE2
> requirement. I'm still fairly convinced that going to i686 is the right
> move - we really don't support i586 as a practical matter, and even
> the Geode should still work with that. Furthermore, it's likely we'll
> have a mass rebuild for LZMA support and/or debuginfo changes, so it's
> no additional cost.