This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Speed up genattrtab
> Hi,
>
> On Wed, 16 Jun 2010, Nathan Froyd wrote:
>
> > > > Pardon the dumb question here, but IIUC, we build insn-attrtab at
> > > > each stage. Why not just build it once for all stages?
> > >
> > > Because that's the definition of bootstrapping. Either we bootstrap,
> > > or we don't. It doesn't make sense to bootstrap only half the
> > > sources.
> >
> > I'm not asking why we compile insn-attrtab at every stage, but why we
> > run genattrtab at every stage. I understand that it's nice to compile
> > genattrtab at each stage for a little sanity checking, but the compiler
> > used to compile genattrtab should have no effect on the output of
> > genattrtab.
>
> If we were going that route, then we wouldn't need bootstrapping at all,
> because the compiler used to compile $random.c should have no effect on
> the output of the final cc1.
>
> > Or were you already responding to "why generate insn-attrtab at every
> > stage"?
>
> Implicitely, yes.
With a lot of cores, we might actually execute getnattrtab and compilation
of stage1 insn-attrtab.c at same time and then just sanity check that genattrtab
did not change results :)
Honza
>
>
> Ciao,
> Michael.