more ia32 bugs (unrolling)

[Horst von Brand]

Toon Moene <toon@moene.indiv.nluug.nl> said:
> Richard Henderson wrote:
> > On Sun, Jun 27, 1999 at 12:12:50PM -0400, Horst von Brand wrote:
> > > > Obviously, max_line is bogus ... However, if you have a moderately
> > > > large system (i.e. like 256 Mbyte), you might just get away with
> > > > it.  > > > > Stop it RIGHT THERE! There are places where 16Mb RAM
> > > > is considered "large", > > > this machine has 32Mb and is average
> > > > around here. Please do _not_ make gcc > > > impossible to build/use
> > > > for large parts of the world.

> > That's not what Toon was suggesting.  Merely that I might not have
> > seen the problem if I was running on a large memory machine.

> It seems that my English is deteriorating [at least in its clarity]. 
> This is the second time in a week that someone has to explain my words
> (the first time was even more embarrassing, as it was meant as a joke -
> see http://www.deja.com , search string "toon dvf comp.lang.fortran")

Sorry, I probably overreacted. But for me it is critical that nobody goes
ahead and considers "256Mb RAM" a "normal machine".

At least not for now ;-)

Thanks all for an outstanding job!

(When 2.95 is out, I'll need some guidance with the memcmp(), memcpy() et
al intrinsecs, the Linux kernel people claim they stink... and I'd love to
purge the reimplementations thereof (cause of much aliasing grief) from the
kernel for good)
-- 
Horst von Brand                             vonbrand@sleipnir.valparaiso.cl
Casilla 9G, Viña del Mar, Chile                               +56 32 672616