This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Improve cost computations in IVOPTS


On Fri, May 29, 2009 at 6:42 PM, Eric Botcazou <ebotcazou@adacore.com> wrote:
>> assuming that compile time and memory consumption are not significantly
>> affected by this (and other similar) changes -- ivopts tries rather hard
>> to avoid folding things and creating new expressions in the phase of the
>> computation of the costs, as I used to run into problems with memory
>> requirements otherwise.
>
> Note that the pass already folds during cost computation (strip_offset_1).
> That's pretty much unavoidable I think, otherwise the discrepancy would be too
> big between the computed costs and the costs of the generated expressions.
>
> I've tested the change on a big Ada application (x86) and on GCC itself
> (x86-64, both individual files and bootstrap), compilation times are not
> affected. ?The binaries are all slightly smaller. ?Memory consumption is
> harder to evaluate but doesn't seem to vary much either for the cases I've
> looked at.
>
> FWIW this brings about measurable improvements on x86 at run time for our
> Ada performance testsuite (from a few percents up to 10 for specific tests).

That's good enough for me.

Thanks for working on this.
Richard.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]