[MPlayer-cygwin] [PATCH] ao directsound

[Diego Biurrun]

Sascha Sommer writes:
> On Wednesday 22 September 2004 12:38, Diego Biurrun wrote:
> > Sascha Sommer writes:
> > > If there are no objections I will commit this tomorrow.
> > > Ao dsound is working a lot better for me than ao win32.
> >
> > Nice to see this, CPU utilization is about the same for me, but I
> > still have a lot of problems, while ao_win32 plays perfectly:

[...]

> > All in all ao_dsound still seems to have some way to go...  This does
> > not mean it should not get committed.  Apparently it works well for
> > many people and it's good if it gets more widespread testing, but I
> > don't think it should be made the default just yet.
> 
> Does it  print something about emulation with -v?

No.

> It would be nice if could make the files at least readable for cvsusers.

Sorry, the samples are world-readable now.

> Maybe commenting out ao_data.outburst = wformat.Format.nBlockAlign * 512; 
> helps?

Unfortunately no.

> Or play with the buffersize.

This does somehow help.  Currently in ao_dsound.c there is

#define BUFFERSIZE 32768

If I change this to

#define BUFFERSIZE 932768

all problems go away.

Nevertheless the values

#define BUFFERSIZE 957504
#define BUFFERSIZE 917504

don't work.  This completely mystifies me.

> However I'm afraid that the problem might be that your soundcard
> returns inaccurate values with
> IDirectSoundBuffer_GetCurrentPosition. Both timing and buffering
> rely on this.

Might well be possible.  I updated the driver to no avail.  Is there
any way to test whether the audio driver is broken?

Diego