This is the mail archive of the
mailing list for the GCC project.
Re: [RFA] attribute((inline_everything)) (was Re: [RFA] Killartificialinlining limit)
- From: Richard Guenther <rguenth at tat dot physik dot uni-tuebingen dot de>
- To: Benjamin Kosnik <bkoz at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Tue, 13 May 2003 18:36:51 +0200 (CEST)
- Subject: Re: [RFA] attribute((inline_everything)) (was Re: [RFA] Killartificialinlining limit)
On Tue, 13 May 2003, Benjamin Kosnik wrote:
> > To get it work, you need to declare your to-be-flat function as both
> > __attribute__((inline_everything)) and __attribute__((noinline)).
> Am I the only one who thinks this is ridiculous? I think the best way
> out of the inlining wars would be
> 1) per-language inlining heuristics, flag controlled
> 2) more accurate estimation of optimized function size
> Hopefully the SSA bits will eventually help with 2.
> Adding all these gross attribute hacks is just going to add clutter and
> hacks that will have to be supported ad infinitum.
Maybe. But if it buys you 1 month computation time out of 3, you still
want to go for it. Its all about speed.
Richard Guenther <richard dot guenther at uni-tuebingen dot de>