Re: [rtcweb] Stephan Wenger's choices

[Stephan Wenger <[e-mail]>]

It's well known that many SoCs (including the Broadcom chip used in Raspberry Pi and some Qualcomm chipsets) include hardware acceleration for some of the VPx codecs.  That's not really a surprise as the low level functional blocks of H.264 and VP8 (and also H.263 and some of the older Vpx codecs, so I hear) are quite similar.  It doesn't take the HW folks that much surface and engineering cycles to add support.
It's also well known that those who put the chips into system designs, in their vast majority, do not enable that acceleration for the decoding of VPx.  Your guess is as good as mine why that is so :-)
It's very hard to get your hand on data sheets etc. of the SoCs, unless you are buying hundreds of thousands of them.  Those who obtained them, cannot talk without violating NDAs.
Stephan


From: Tim Panton <[e-mail]<mailto:[e-mail]>>
Date: Mon, 30 Dec 2013 19:47:26 +0000
To: Eric Rescorla <[e-mail]<mailto:[e-mail]>>
Cc: "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>
Subject: Re: [rtcweb] Stephan Wenger's choices


On 30 Dec 2013, at 18:33, Eric Rescorla <[e-mail]<mailto:[e-mail]>> wrote:

2. Mobile devices on which we want to deploy WebRTC but which can't
afford to run a modern codec in software but could perhaps use HW
support (and perhaps could run some older codec).

...
The second category of elements would benefit from H.264 in the
sense that they would be able to play video or have better battery
life or whatever to the extent to which hardware is available and
usable. The extent to which that's a benefit depends on how often
that will be the case; I realize that a lot of people believe that will
be rare, but the evidence I have seen so far seems pretty
equivocal. Sure, if you think you will never be able to use
HW acceleration for H.264, then this doesn't seem like a very
good argument.

There is a subtle assumption there that hardware assist only applies to H264.
I notice that the raspberryPi gained experimental VP8 support using the video GPU
to accelerate it.
http://www.cnx-software.com/2013/01/26/raspberry-pi-now-has-experimental-support-for-vp6-vp8-mjpeg-and-ogg-theora-video-codecs/

What is especially interesting is that this didn't require any hardware changes,
so it may be that many existing 'H264 only' devices could follow suit if there was
a suitable motivation.

Tim.

_______________________________________________ rtcweb mailing list [e-mail]<mailto:[e-mail]> https://www.ietf.org/mailman/listinfo/rtcweb