Speed up genattrtab
Richard Guenther
richard.guenther@gmail.com
Wed Jun 16 09:32:00 GMT 2010
On Tue, Jun 15, 2010 at 8:39 PM, Mark Mitchell <mark@codesourcery.com> wrote:
> Jan Hubicka wrote:
>
>> In WHOPR compilation we now partition the program based on source files (this
>> is something I plan to revisit later), so we get partition that corresponds to
>> insn-attrtab. WPA driver sorts units by size and feeds them to parallel make
>> and insn-attrtab is still last on finishing build by quite long shot. This
>> ignore time spent by genattrtab itself.
>>
>> So just reordering makefiles is not going to solve the problem.
>
> Well, it might solve the problem for developers not using WHOPR. :-)
>
> I certainly don't claim it's a general solution; at some point, if you
> have enouh cores, the single core compiling insn-attrtab will dominate
> anyhow. But if some easy Makefile hackery would help some developers,
> I'd still argue we should do it. And I still have a hard time with a
> change that's going to make the compiler slower in order to save a
> minute or two during compiler builds, especially given that test cycles
> are so much longer than build cycles.
I think the makefile ordering is already good - it is just that genattrtab
takes so much time that compiling insn-attrtab starts very late. Michael
addresses this by speeding up genattrtab.
Richard.
> --
> Mark Mitchell
> CodeSourcery
> mark@codesourcery.com
> (650) 331-3385 x713
>
More information about the Gcc-patches
mailing list