This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH,spu]: generate inline code for int-to-double conversion
- From: Sa Liu <SALIU at de dot ibm dot com>
- To: "Andrew Pinski" <pinskia at gmail dot com>
- Cc: Andrew_Pinski at playstation dot sony dot com, gcc-patches at gcc dot gnu dot org, trevor_smigiel at playstation dot sony dot com
- Date: Wed, 19 Sep 2007 11:42:29 +0200
- Subject: Re: [PATCH,spu]: generate inline code for int-to-double conversion
Ping?
"Andrew Pinski" <pinskia@gmail.com> wrote on 13.09.2007 12:55:19:
> On 9/13/07, Sa Liu <SALIU@de.ibm.com> wrote:
> > The current spu-gcc compiler always calls libgcc functions
float_unssidf
> > and float_unsdidf when converting integer to double. This reduces code
> > size, but costs performance.
> >
> > This patch is about to generates inline code for the conversion by
> > default, and switchs on the libcall only if -Os is given.
> > + rtx c0 = spu_const_from_ints(V16QImode, 0x02031011, 0x12138080,
> > + 0x06071415, 0x16178080);
> > + rtx r0 = gen_reg_rtx(V16QImode);
>
> You need a space between the function name and the left parentheses.
> I have not looked into the rest of the patch (yet).
>
> Thanks,
> Andrew Pinski