[MPlayer-users] Issues with -vc ff*vdpau and -vc ff*vdpauold

[Reimar Döffinger]

On Sat, Nov 01, 2014 at 03:52:31PM +0100, wm4 wrote:
> On Sat, 1 Nov 2014 15:34:57 +0100
> Reimar Döffinger <Reimar.Doeffinger at gmx.de> wrote:
> > For regression testing it is however sufficient to hash the critical
> > structs passed to the VDPAU API.
> > I think I've posted patches for that already, but it was a bit of a
> > mess.
> > However what I'd really like but absolutely have no time for is a
> > software reference implementation of the VDPAU API.
> > This would allow testing on all FATE machines, regardless of what
> > hardware they have.
> > (as a first step such a reference implementation could also just
> > serialize out the API calls up to that frame, which would be
> > pure regression testing then).
> 
> There's already a "tracing" pseudo-driver [1] provided by nvidia. Would
> that be helpful?
> 
> [1] http://cgit.freedesktop.org/~aplattner/libvdpau/tree/trace/vdpau_trace.cpp

I have not yet looked enough to know how to actually use it.
It might be a good starting point, but it only offers the choice of
printing the data buffers fully or no contents at all.
For regression testing we'd probably want hashes instead.
Also that tracing implementation still needs an actual implementation
behind it, so it might help for regression testing with hardware,
but not if you do not have hardware.