[FFmpeg-devel] [PATCH]Remove xvmc_acceleration if favor of hw_acceleration

Reimar Döffinger Reimar.Doeffinger
Thu Jan 8 15:41:32 CET 2009


On Thu, Jan 08, 2009 at 03:21:24PM +0200, Ivan Kalvachev wrote:
> On 1/7/09, Reimar D?ffinger <Reimar.Doeffinger at stud.uni-karlsruhe.de> wrote:
> > Why is that an issue? Because it gets confusing as hell if someone
> > actually does add hardware decoding, do we name the next enum then
> 
> And what do you think the actual hardware decoding would be doing?

Actually decode? I can see why it was named like that earlier but the
assumption that hardware decoding will always go directly to the display
and thus "hardware decoding" means the actual decoding will be in the
display code is no longer justified IMO.

> Or rather, do you advertise that the current model is wrong and
> libavcode should be calling the hardware acceleration API directly?

None of the implementations we have are suitable for that, they all
decode for display only. Hardware accelerated decoding (that actually
does decode within FFmpeg) would have to decode into system memory again.

> It definitely could save some duplication of code among the players.

Hardly, it would not be suitable for most players since the performance
of decoding to system memory would be far worse.

Greetings,
Reimar D?ffinger




More information about the ffmpeg-devel mailing list