[Theora-dev] Questions, MMX and co.

[Rodolphe Ortalo]

On Thursday 26 August 2004 01:43, Ralph Giles wrote:
[...]
> All of them? :) GCC and MSVC at a minimum.
[...]
> You forgot raw .asm files. Concensus on irc was both. the intrinsics are
> easier to understand and can help take advantage of compiler
> improvements for scheduling; raw assembly is about the only thing you
> can make portable.

Does it mean you would rule out inline assembly altogether?

> > 4) How to benchmark the implementation? (I'm still using the small
> > wav+yuv video with this cute little girl singing but I guess something
> > more serious should be done...) If possible, it should be easily
> > accessible to everyone (no expensive digital equipment, no multi-gigabyte
> > downloads) so that everyone could reproduce the test and compare results.
>
> Source is always going to mean multi-gigabyte downloads. We have a
> collection, but some of it needs to be put back online. Obviously with
> decoder benchmarks it's a little easier.

I'm playing with commercial DVDs (original ones). e.g. for the first chapter 
of Lord of the Rings - 2 towers (long edition - french edition of course but 
I guess the audio difference is not very important in the encoding time), I 
obtained execution times like these (user time):
C-code (gcc-3.4 -O6) 9m17s
MMXEXT inline (gcc-3.4 -O6) 4m57s
some MMXEXT intrinsics (gcc-3.4 -O6 -mmmx -msse) 5m17s

Needed HD space is not so important (between 20Mo or 150Mo ogg output).
I find this approach is not so bad finally. I can post my shell scripts for 
doing the whole encoding if there is interest.
(Of course, if sci-fi/fantasy is banned we can shift to Amelie Poulain... :-)

> > 6) Who has the definitive answer on the above questions: or btw, who
> > rules Xiph/Theora? :-)
>
> Me.

ACK. At least one point is answered. :-)

Rodolphe