[Advocacy] persuading mobile phone manufacturers to implement Vorbis support

[xiphmont at xiph.org]

On 5/29/07, Ivo Emanuel Gonçalves <justivo at gmail.com> wrote:
> On 5/28/07, Ian Malone <ibmalone at gmail.com> wrote:
> > However several Symbian based phones can run Vorbis
> > through third party software, this isn't the obstacle
> > in those cases.
>
> Yes.  What Monty meant was that old hardware didn't have enough power
> to run Vorbis, so MP3 became the winner in spite of its bad quality,
> non-standard design and expensive license fees.

This is a bit off in a few ways.  To begin, MP3 predates Vorbis.  MP3
hadn't hit the peak of its popularity yet but it was already the clear
defacto standard; it had 'won' before Vorbis existed. The original
point of Vorbis was to provide a free codec alternative to MP3 when
Fraunhofer et al. opted to shut down all the free software projects
that had given MP3 its foothoold to begin with.

Second, virtually any DSP has the CPU power to handle Vorbis.  It is
entirely a question of working memory; very few DSPs have more than a
few kilowords of available working memory and the smallest encoder
footprint I've personally coded required about 20kW.

At the time Vorbis began, several new Cirrus ARM-based DSPs (the
Maverick series) were really taking off, and the most common
applications of the Maverick used offboard DRAm, ie, alot of working
memory (the onboard SRAM was used as a cache).  Other manufacturers
followed suit.  This looked like it was going the be the overwhelming
trend in handheld players... and in a way it was.  This is the
strategy used by iPod, Zune and most cellphones.  But it did not turn
out to be a large enough majority of players.

It is not ARM specifically that Vorbis requires, it is the fact that
ARM-based players tend to have alot of useful RAM.

Monty