This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH, SPU] Improved __divdf3 routine
- From: trevor_smigiel at playstation dot sony dot com
- To: Ulrich Weigand <uweigand at de dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org, andrew_pinski at playstation dot sony dot com
- Date: Wed, 25 Mar 2009 14:18:06 -0700
- Subject: Re: [PATCH, SPU] Improved __divdf3 routine
- References: <200903252028.n2PKSfXS022398@d12av02.megacenter.de.ibm.com>
Hi Ulrich,
Yes, this is OK for mainline.
Trevor
* Ulrich Weigand <uweigand@de.ibm.com> [2009-03-25 13:35]:
> Hello,
>
> this patch adds an optimized version of the divdf3 routine for
> the SPU. It is fully vectorized, but even if used just for scalar
> operations, it is still more than twice as fast than the default
> routine. In addition, it supports non-default rounding modes a lot
> better than the fp-bit.c implementation.
>
> The implementation is taken from Dan Brokenshire's most current
> version of the SIMDmath library code.
>
> Tested on spu-elf with no regressions.
>
> OK for mainline?
>
> Bye,
> Ulrich
>