Re: [OT] Disabling hardware speculation support in GCC flags
On Sun, 7 Dec 2008 12:25:03 +0530
"Sujit Karataparambil" <sjt.kar@gmail.com> wrote:
> > You're right, I guess we can't disable the "hardware" directly but if
> > we let gcc know about this then it is possible that it will do further
> > optimizations correct?
> What sort of run time issues can addressed on an runtime usage.
>
I'm sorry I am not sure what you mean by this
> > For example, I can compile a C program using regular gcc without any
> > options and then adding the following and comparing the differences:
> >
> > "-mcpu=7400 -O2 -pipe -maltivec -mabi=altivec"
>
> These are stricly static rules.
>
> >
> > This way the compiler "knows" about the hardware and tries to optimize
> > the program based on that.
>
> How will you make an runtime sequence be made correclty to an compiler
> >
> > What do you think?
>
> reminds me of hennesey and patternson chapter 03.
>
Yeah you have good memory. I am reading that book and trying to better
understand the difference between compiler speculation and hardware
speculation.
> >
> > Thanks,
> > Amit
> >
> >
> > --
> > To UNSUBSCRIBE, email to debian-powerpc-REQUEST@lists.debian.org
> > with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
> >
> >
>
>
> --
> --linux(2.4/2.6),bsd(4.5.x+),solaris(2.5+)
Reply to: