--- Abhijit Bhopatkar <[EMAIL PROTECTED]> wrote:
> A question. is pipelining dependant on the code? i mean
> why would one
> need to turn on optimisations? won't the i686 pipeline
> work even otherwise?
I dont think its possible to "turn off" the pipeline - even
if he's forced a switch to virtual 8086 mode.
I think what he means by "turning on the optimizations" is
to *not* use one of GCC's optimization flags - some of
which are highly architecture dependant.
>
> Thess are quite stupid comments actually, but i was
> really wondering if
> anyone would turn off the opt. in any case, except for
> may be
> portability advantage.
Hmmm interesting point. I do suppose that when we put in
all the various optimization flags to gcc, we are
implicitely restricting our intended platforms.
On the other hand, -funrollloops should not have any effect
on the portability at all - Its just making an assumption
that the pipeline will like the resulting machine code.
The only guranteed way to restrict portability is to use
assembly language and then its still usable on a range of
x86.
And if you really wanted that your code be portable (or
non-portable - as you see it), then masm and nasm both have
a ".$x86" directive that allows instructions only upto your
choice of $x.
But I dont know any way in which you'd be able to tell gcc
something like this.
Anyone who knows whether something like this exists?
Uv
___________________________________________________________
How much free photo storage do you get? Store your holiday
snaps for FREE with Yahoo! Photos http://uk.photos.yahoo.com
--
______________________________________________________________________
Pune GNU/Linux Users Group Mailing List: (plug-mail@plug.org.in)
List Information: http://plug.org.in/mailing-list/listinfo/plug-mail
Send 'help' to [EMAIL PROTECTED] for mailing instructions.
