This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
RE: [patch] Fixed-point patch 8/10
- From: "Fu, Chao-Ying" <fu at mips dot com>
- To: "Richard Sandiford" <richard at codesourcery dot com>, <gcc-patches at gcc dot gnu dot org>
- Cc: "Mark Mitchell" <mark at codesourcery dot com>, "Thekkath, Radhika" <radhika at mips dot com>, "Stephens, Nigel" <nigel at mips dot com>
- Date: Mon, 10 Sep 2007 12:03:40 -0700
- Subject: RE: [patch] Fixed-point patch 8/10
Richard Sandiford wrote:
>
> My first thought was "why not do this in target-independent code?".
> I.e., if the target doesn't have an addMM3 pattern for some MM for
> which overflow is undefined, why not try generating ssaddMM3 or
> usaddMM3 instead, before falling back on an element-wise
> implementation?
> Presumably the trick will be useful for other targets too. Or is the
> idea that, by keeping the rtl code as a PLUS rather than [US]S_PLUS,
> later optimisations might be able to take advantage of the
> undefinedness?
>
> I'd like to know your thoughts and others' thoughts on that first.
> If we do decide that it's the target's responsibility to duplicate
> the patterns, the patch looks reasonable.
>
> I think this is now the first uncommitted patch in the series.
> FWIW, the approval for the original patch still stands, so feel
> free to commit that version now if you like, and deal with this
> as a follow-on. If you'd prefer to get the extra patterns sorted
> out first, that's fine too.
>
>
Ok. I will check in the original version first. People can discuss
and decide which way is better for the new stuff later. Thanks!
Regards,
Chao-ying