[FFmpeg-devel] [PATCH] split-radix FFT

Michael Niedermayer michaelni
Tue Jul 29 16:46:22 CEST 2008


On Tue, Jul 29, 2008 at 05:32:44PM +0300, Uoti Urpala wrote:
> On Tue, 2008-07-29 at 16:18 +0200, Michael Niedermayer wrote:
> > On Tue, Jul 29, 2008 at 07:36:06AM -0600, Loren Merritt wrote:
> > > On Tue, 29 Jul 2008, Michael Niedermayer wrote:
> > > >
> > > > Also if i use MANGLE to access the 2 root2 globals things clear up a bit in
> > > > the code generated, so i would suggest that to be done at least, even if
> > > > the asm stays split up.
> > > 
> > > If we don't actually require the output to be position independent, isn't 
> > > the simpler solution to not compile with -fPIC? Or are we doing some sort 
> > > of informed trade-off where only the most cpu-intensive functions get 
> > > textrels?
> > 
> > Thats a good argument as well.
> 
> Funny how you didn't find the argument good at all when I mentioned it
> while talking about MANGLE earlier... :)

You argued in favor of using MANGLE ?
I distinctly remember you arguing in favor of replacing all MANGLE by
"m". Which really is the opposite ...


> 
> >  Though i was more thinking of distributions
> > that have a policy that says -fPIC must be added to the flags.
> > 
> > http://www.debian.org/doc/debian-policy/ch-files.html:
> > 10.2 Libraries
> 
> Currently FFmpeg in Debian is built without -fPIC on x86.

Iam happy to hear this, still the point stands, distributions have such
requirements. And it takes the package maintainer quite a bit of work (or
rulebreaking) to not compile with -fpic.

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The worst form of inequality is to try to make unequal things equal.
-- Aristotle
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20080729/77389b0c/attachment.pgp>



More information about the ffmpeg-devel mailing list